JP2016521966A - Compositions and methods for cancer prognosis - Google Patents

Abstract

Provided herein are methods for assessing cancer samples from patients, such as prostate tumor samples, such as computer-implemented or automated methods. A biomarker panel for prognosis of prostate cancer is also provided. Also provided are methods of treating prostate cancer by identifying prostate cancer that may have aggressive prostate cancer or a fatal outcome. The method comprises 1, 2, 3, 4, 5, 6, 7 or 8 tumor markers of FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set) or said tumor marker Identifying the level, eg, amount or expression level, of one or more DNAs or mRNAs, thereby assessing the tumor sample.

Description

This application claims the benefit of US Provisional Application No. 61 / 792,003, filed Mar. 15, 2013, the entire contents of which are hereby incorporated by reference in their entirety.

The present invention relates to the use of a biomarker panel to predict the prognosis of cancer patients.

BACKGROUND OF THE INVENTION Prostate cancer (PCA) is the most common cancer in men. Most older men have intraprostatic neoplasia, but the overwhelming majority of cases remain local and indolent, and there is no need for therapeutic intervention. However, there is a subset of early stage PCA in which aggressive malignancies are “hard to change” and, if left untreated, spread outside the prostate and progress mercilessly to metastatic disease and ultimately death To do. Because of the inability to accurately discriminate between indolent and aggressive diseases, many men with potentially indolent diseases now have unnecessary radical therapeutic interventions with high unhealthy rates, such as prostatectomy and Receiving radiation. In the United States alone, the costs associated with prostate cancer overtreatment are estimated to exceed $ 2 billion annually. This also does not include the impact on the quality of life of the therapeutic treatment. On the other hand, some patients with potentially aggressive PCA are inadequately treated and die due to disease progression.

  The stratification methods currently used to predict prostate cancer outcomes are based on clinical factors such as Gleason classification, prostate specific antigen (PSA) levels and tumor progression classification. However, these factors are not well-predicted and are not reliably associated with the most important clinical endpoints, metastatic risk and PCA-specific mortality. With this medical need not yet addressed, the genetic and biological basis for PCA progression with the goal of identifying biomarkers that can be assigned progression risk and provide opportunities for targeted interventional treatment Efforts to prescribe are invigorating. Genetic studies of human PCA have led to several indicator events such as inactivation of PTEN tumor suppressor and ETS family translocation and dysregulation as well as other genetic or epigenetic modifications such as Nkx3.1, c-Myc and SPINK Has been identified. Global molecular analysis has also identified a series of potential relapse / metastasis biomarkers such as ECAD, AIPC, Pim-1 Kinase, hepsin, AMACR and EZH2. However, the extremely strong heterogeneity of human PCA limits the use of single biomarkers in clinical settings, and therefore a more comprehensive transcription profiling study to define a multigene biomarker panel or indicator for prognosis Is inspired. Although such panels or indicators appear to be more robust, their clinical use has been linked to the myriad bystander genetic and epigenetics that lead to the inherent noise and context-specific nature of transcriptional networks and significant disease heterogeneity. Remains uncertain due to the extreme instability of the cancer genome with genetic events. Thus, there is a need for a more accurate prognostic test in early stage tumors that can be used to predict the presence and behavior of cancer, particularly in early stages, and thus useful in guiding appropriate treatment for prostate cancer patients. ing.

SUMMARY OF THE INVENTION In one aspect, provided herein is a method for assessing a cancer sample, eg, a prostate tumor sample, from a patient, eg, a computer-implemented method or an automated method. The method comprises 1, 2, 3, 4, 5, 6, 7 or 8 tumor markers of FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set) or said tumor marker Identifying the level, eg, amount or expression level, of one or more DNAs or mRNAs, thereby assessing the tumor sample.

  In embodiments, the method comprises obtaining a tumor marker signal, for example, directly or indirectly. In embodiments, the method includes obtaining the signal directly.

  In embodiments, the method includes obtaining a cancer sample directly or indirectly.

  Also, in the present specification, 1, 2, 3, 4, 5, of (a) cancer sample; and (b) FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set). There is provided a reaction mixture comprising 6, 7 or 8 kinds of tumor markers or reagents for detecting DNA or mRNA of the tumor markers. In some embodiments of the reaction mixture, the cancer sample comprises a plurality of portions, such as slices or aliquots. In some embodiments of the reaction mixture, the first portion of the cancer sample comprises a detection reagent for the first marker, but not all of the markers, and the second portion of the cancer sample comprises the first The detection reagent for one of the markers that does not contain the detection reagent for the other marker is included.

  Also provided herein are methods for assessing patient samples, eg, tissue samples, eg, cancer samples, eg, prostate tumor samples, eg, computer-implemented methods or automated methods. The method includes: (a) identifying, from the sample, a level of a phenotypic marker of a first region, eg, a first tumor marker, in a region of interest (ROI), thereby evaluating the sample. It is a waste.

  In embodiments, the sample is a cancer sample. In embodiments, the sample comprises solid tumor-derived cells. In embodiments, the sample comprises cells from a humoral tumor. In embodiments, the ROI is defined or selected by morphological features.

  In embodiments, the ROI is defined by manual or automated means and physical separation of the ROI from other cells or materials, eg, by incision of the ROI, eg, cancerous area, from other tissues, eg, non-cancerous cells. Or selected. In embodiments, the ROI is defined or selected by non-morphological features, such as ROI markers. In embodiments, ROIs are identified or selected by cell sorting by including ROI markers. In embodiments, the ROI is identified or selected by a combination of morphological and non-morphological selection.

  In embodiments, the level of a phenotypic marker of a first region, eg, a first tumor marker, is identified in a first ROI, eg, a first cancerous region, and a phenotypic marker of a second region, eg, a first Two tumor marker levels are identified in a second ROI, eg, a second cancerous region.

  In embodiments, the phenotypic marker level of the first region and the phenotypic marker level of the second region, eg, the tumor marker level, are identified in the same ROI, eg, the same cancerous region.

  In embodiments, the method further comprises: (b) identifying an ROI, eg, an ROI corresponding to a cancerous region.

  In some embodiments, (a) is performed before (b).

  In other embodiments, (b) is performed before (a).

  In embodiments, the identification of the level of a first region phenotypic marker, eg, a first tumor marker, is associated with binding of a detection reagent to said first region phenotypic marker, eg, a first tumor marker. For example, obtaining a proportional signal, eg directly or indirectly.

  In embodiments, the method comprises contacting the sample with a first region phenotypic marker, eg, a detection reagent for the first tumor marker.

  In embodiments, the method comprises contacting the sample with an ROI marker, eg, an epithelial marker detection reagent.

  In embodiments, the method further includes obtaining an image of the sample and analyzing the image. In some such embodiments, the method includes calculating the patient's risk score from the image.

  In embodiments, the method comprises contacting the sample with a phenotypic marker of a first region, eg, a tumor marker detection reagent, and obtaining a detection reagent binding value. In some such embodiments, the method includes calculating the patient's risk score from the value.

  In embodiments, the method further comprises (b) contacting the sample with a ROI marker detection reagent. In embodiments, the method further includes (c) defining an ROI. In embodiments, the method further comprises (d) identifying the level of a region-phenotype marker, eg, a tumor marker, in the ROI. In embodiments, the method further comprises (e) analyzing the level to obtain a risk score. In embodiments, the method further includes repeating steps (a)-(d).

  In embodiments, the method further comprises (i) subjecting the sample to one or more physical preparation steps of the sample, eg, dissociating the sample, eg, trypsinizing, incising the sample. Or (ii) contacting the ROI with a detection reagent; and / or (iii) detecting a signal from the ROI.

  The present invention also relates to a method of evaluation of a patient tumor sample, such as a prostate tumor sample, for example a computer-implemented method or an automated method:

(A) In a ROI, eg, a cancerous ROI, a first region-phenotype marker, eg, a first tumor marker (eg, the first tumor marker is FUS, SMAD4, DERL1, YBX1, pS6, PDSS2 , Selected from CUL2 and HSPA9 (tumor marker set)) or the DNA or mRNA level, eg, amount, of said first tumor marker, thereby assessing said tumor sample To do.

  In embodiments, the level of a first region-phenotype marker, eg, a first tumor marker from said tumor marker set, is identified in a first ROI, eg, a cancerous ROI, and a second region-phenotype marker For example, the level of a second tumor marker from the tumor marker set is identified in a second ROI, such as a second cancerous ROI. In embodiments, the first ROI, eg, cancerous ROI, and the ROI, eg, second cancerous ROI, are identified or selected by the same method or criteria. In embodiments, the first region-phenotypic marker level and the second region-phenotypic marker level (e.g., both first and second tumor markers from the tumor marker set) within the same ROI, For example, identified within the same cancerous ROI.

  In embodiments, the method further comprises: (b) identifying the ROI, eg, the ROI of the tumor sample corresponding to the tumor epithelium. In some embodiments of the method, (a) is performed before (b). In some embodiments of the method, (b) is performed before (a).

  In embodiments, the identification of the level of a first region-phenotype marker, eg, a first tumor marker, is associated with binding of a detection reagent to said first region-phenotype marker, eg, a first tumor marker. For example, obtaining a proportional signal, eg directly or indirectly. In embodiments, the tumor marker is DNA encoding FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2, or HSPA9HSPA9. In embodiments, the tumor marker is mRNA encoding FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. In embodiments, the tumor marker is a protein selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9.

  In embodiments, the method comprises contacting a sample with a detection reagent for a marker in a tumor marker set, obtaining an image of the sample directly or indirectly, and analyzing the image. In embodiments, the method includes calculating a patient risk score from the image.

  In embodiments, the method comprises contacting a sample with a detection reagent of a first marker of a tumor marker set, and obtaining a detection reagent binding value directly or indirectly. In embodiments, the method includes calculating the patient's risk score from the value.

  Any of any of the foregoing methods In one embodiment, the method is further selected from the set of tumor markers in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying a second tumor marker or DNA or mRNA level of said second tumor marker.

  In embodiments, the second tumor marker is a protein of the tumor marker set.

  In embodiments, the method further comprises a third tumor marker selected from the tumor marker set or of the third tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises a fourth tumor marker selected from the set of tumor markers or the fourth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises a fifth tumor marker selected from the tumor marker set or the fifth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises a sixth tumor marker selected from the tumor marker set or the sixth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises a seventh tumor marker selected from the tumor marker set or of the seventh tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises an eighth tumor marker selected from the tumor marker set or an eighth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. Identifying the level of DNA or mRNA.

  In embodiments, the method further comprises identifying the level of additional markers disclosed herein other than those of the tumor marker set.

  In embodiments, the level of said additional marker is identified in a cancerous ROI.

  In embodiments, the level of the additional marker is identified in a benign ROI.

  In embodiments of any one of the foregoing methods, the method further comprises providing the tumor sample or the cancer sample. (As used herein, the terms “cancer sample” and “tumor sample” are interchangeable unless indicated otherwise in the text.)

  In embodiments of any one of the foregoing methods, the method further comprises the tumor sample from another entity, such as a hospital, laboratory, or clinic.

  In any one embodiment of the foregoing method, the cancer sample or the tumor sample comprises a section or slice of prostate tissue.

  In any one embodiment of the foregoing method, the cancer sample or the tumor sample comprises a plurality of portions, eg, a plurality of prostate tissue sections or slices.

  In any one embodiment of the foregoing method, the cancer sample or the tumor sample is fixed, eg, formalin fixed.

  In embodiments of any one of the foregoing methods, the cancer sample or the tumor sample is embedded in a matrix.

  In embodiments of any one of the foregoing methods, the cancer sample or the tumor sample is embedded in paraffin.

In embodiments of any one of the foregoing methods, the cancer sample or the tumor sample is deparaffinized.

  In any one embodiment of the foregoing method, the cancer sample or the tumor sample is a formalin-fixed paraffin embedded sample or an equivalent thereof.

  In embodiments, cancer sample or tumor sample preparation (eg, deparaffinization) is automated.

  In embodiments of any one of the foregoing methods, contact of the detection reagent with the cancer sample or tumor sample is automated.

  In embodiments of any one of the foregoing methods, the cancer sample or tumor sample is placed in an automated scanner.

  In embodiments of any one of the foregoing methods, a cancer sample or tumor sample, eg, a portion of prostate tissue, eg, a section or slice, is placed on a substrate, eg, a solid or rigid substrate, eg, glass or plastic. Place on substrate, eg glass slide. In some such embodiments, a first portion of the tumor sample, such as a section or slice, is placed on a first substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. Put it on. In embodiments, a second portion of the tumor sample, such as a section or slice, is placed on a second substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. In embodiments, a third portion of the tumor sample, such as a section or slice, is placed on a third substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. In embodiments, a fourth portion of the tumor sample, such as a section or slice, is placed on a fourth substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide.

  In embodiments, the first and second portions are analyzed simultaneously. In embodiments, the first and second portions are sequentially analyzed.

  In embodiments of any one of the foregoing methods, the detection reagent comprises a tumor marker antibody, eg, a tumor marker monoclonal antibody, eg, a tumor marker antibody against FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. . In embodiments, the tumor marker antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

  In embodiments, the detection reagent comprises a second antibody that is an antibody to the tumor marker antibody, eg, a monoclonal antibody.

  In embodiments, the detection reagent comprises a third antibody that is an antibody against the second antibody, eg, a monoclonal antibody.

  In embodiments, the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

  In embodiments, the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

In embodiments of any one of the foregoing methods, the cancer or tumor sample is:
A first ROI marker detection reagent, eg, a whole epithelial detection reagent (eg, as described herein), having a first luminescence profile, eg, a first peak emission, or measured in a first channel ;
A second ROI marker detection reagent, eg, a basal epithelial detection reagent (eg, as described herein) having a second emission profile, eg, a second peak emission, or measured in a second channel ;
A region-phenotype marker, eg, a tumor marker detection reagent (eg, as described herein) having a third emission profile, eg, a third peak emission, or measured in a third channel
Contact with.

  In embodiments, the cancer or tumor sample is further contacted with a nuclear detection reagent having a fourth luminescence profile, eg, a fourth peak luminescence, or measured in the fourth channel.

  In embodiments, a sample of cancer or tumor is further treated with a second region-phenotype marker, such as a first luminescence profile, eg, having a fifth peak emission, or measured in a fifth channel. Contact with two tumor marker detection reagents (eg, those described herein).

  In embodiments, a sample of cancer or tumor is further treated with a third region-phenotypic marker, such as a sixth luminescence profile, eg, a sixth peak luminescence, or measured in a sixth channel. 3. Contact with 3 tumor marker detection reagents (eg, those described herein).

  In embodiments of any one of the foregoing methods, the identification of an ROI, eg, a cancerous ROI, includes identifying a region having an epithelial structure without the outer layer of basal cells.

  In embodiments, the epithelial structure is detected with a first ROI-specific detection reagent, eg, a first whole epithelial-specific detection reagent, eg, an antibody, eg, a monoclonal antibody, eg, an anti-CK8 or anti-CK18 antibody, eg, a monoclonal antibody. Is done.

  In embodiments, the epithelial structure comprises the first ROI specific detection reagent, eg, the first total epithelial specific detection reagent and the second ROI specific detection reagent, eg, the second total epithelial specific detection. Detected with reagent.

  In embodiments, one of the first ROI-specific detection reagent, eg, the first total epithelial-specific detection reagent and the second ROI-specific detection reagent, eg, the second total epithelial-specific detection reagent Is a CK8 detection reagent, eg, an anti-CK8 antibody, eg, a monoclonal antibody, and the other is a CK18 binding reagent, eg, an anti-CK18 antibody, eg, a monoclonal antibody.

  In embodiments, the binding signal of the first ROI-specific detection reagent, eg, the first whole epithelial detection reagent, is detected through the first channel, eg, at a first wavelength.

  In embodiments, the binding signal of the first ROI-specific detection reagent, eg, the first total epithelial detection reagent, and the second ROI-specific detection reagent, eg, the signal of the second total epithelial detection reagent Is detected at the first wavelength, for example, via the first channel.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial detection reagent, comprises a marker antibody, eg, a marker monoclonal antibody.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial detection reagent, is conjugated to a label, eg, a fluorescent moiety, eg, a fluorescent dye. Has been.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial binding agent, is an antibody against the marker antibody, eg, a monoclonal antibody. 2 antibodies.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial binding agent, is an antibody to the second antibody, eg, a monoclonal antibody. Contains a third antibody.

  In embodiments, the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

  In embodiments, the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

  In embodiments, the presence or absence of basal cells is detected with a ROI-specific detection reagent, such as a basal epithelial detection reagent, such as the basal epithelial detection reagent described herein.

  In embodiments, the method further comprises identifying a ROI corresponding to the benign ROI of the tumor sample, eg, a second ROI.

  In embodiments, identifying a benign ROI includes identifying a region having an epithelial structure surrounded by an outer layer of basal cells.

  In embodiments, basal cells are detected with a ROI-specific detection reagent of the basal epithelium, such as an antibody, such as a monoclonal antibody, such as an anti-CK5 antibody, such as a monoclonal antibody or an anti-TRIM29 antibody, such as a monoclonal antibody.

  In embodiments, the basal cells comprise said ROI-specific detection reagent for basal epithelium and a second ROI-specific detection reagent for basal epithelium, such as an antibody, such as a monoclonal antibody, such as an anti-CK5 antibody, such as a monoclonal antibody or an anti-TRIM29 antibody. For example, it is detected with a monoclonal antibody.

  In embodiments, one of the first ROI-specific detection reagent for basal epithelium and the ROI-specific detection reagent for basal epithelium is a CK5 detection reagent, such as an anti-CK5 antibody, such as a monoclonal antibody, and the other is a TRIM29 detection reagent. For example, an anti-TRIM29 antibody, such as a monoclonal antibody.

  In embodiments, the binding signal of the first ROI-specific detection reagent in the basal epithelium is detected via the first channel, for example at a first wavelength.

  In embodiments, the signal of the binding of the first ROI-specific detection reagent of the basal epithelium and the signal of the second ROI-specific detection reagent of the basal epithelium are transmitted through the first channel, for example, at a first wavelength. Is detected.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent of the basal epithelium comprises a marker antibody, eg, a marker monoclonal antibody.

  In embodiments, the first (and optionally second, if present) ROI-specific detection reagent of the basal epithelium is conjugated to a label, eg, a fluorescent moiety, eg, a fluorescent dye.

  In embodiments, the first (and optionally the second, if present) ROI-specific detection reagent of the basal epithelium comprises a second antibody, eg, a monoclonal antibody, against the marker antibody.

  In embodiments, a third antibody wherein the first (and optionally the second, if present) ROI-specific detection reagent of the basal epithelium is an antibody against the second antibody, eg, a monoclonal antibody including.

  In embodiments, the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

In embodiments, the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye.

  In embodiments, the method further comprises identifying the ROI of the tumor sample as a stroma.

  In embodiments of any one of the foregoing methods, the method comprises (i.a) obtaining a signal of a whole epithelial specific marker, eg CK8, directly or indirectly; (ii.a) basal epithelial specific It involves obtaining the signal of a marker, eg CK5, directly or indirectly.

In embodiments of any one of the foregoing methods, the method further comprises: (ib) obtaining a signal of a second whole epithelial specific marker, eg CK18, directly or indirectly; (ii. b) obtaining directly or indirectly a signal of a second basal epithelial specific marker, eg TRIM29. In embodiments, the method further includes (iii) obtaining a nuclear marker signal directly or indirectly. In embodiments, the method further comprises (iv) obtaining a signal of a second tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (v) obtaining a signal of a third tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (vi) obtaining a signal of a fourth tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (vii) obtaining a signal of a fifth tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (viii) obtaining a signal of a sixth tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (ix) obtaining a signal of a seventh tumor marker of the tumor marker set directly or indirectly. In embodiments, the method further comprises (x) obtaining the signal of the eighth tumor marker of the tumor marker set directly or indirectly.

  In embodiments, the signals of (ia) and (ib) have the same peak emission or are collected in the same channel.

  In embodiments, the signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel.

  In embodiments of any one of the foregoing methods, the method comprises: (i.a) obtaining a signal of a whole epithelial specific marker, such as CK8, directly or indirectly; (ib) a second Obtaining directly or indirectly a signal of a whole epithelial specific marker such as CK18; (ii.a) obtaining a signal of a basal epithelial specific marker such as CK5 directly or indirectly; (ii.b) Obtaining a signal of a second basal epithelial specific marker, for example TRIM29, directly or indirectly; (iii) obtaining a signal of a nuclear marker directly or indirectly; (iv) a signal of a first tumor marker (V) obtaining the signal of the second tumor marker directly or indirectly; or (vi) the third tumor marker. It comprises obtaining a signal directly or indirectly. In embodiments, the method comprises (i.a), (ii.a), (iii) and (iv). In embodiments, the method comprises (ia), (ib), (ii.a), (ii.b), (iii) and (iv). In embodiments, the method includes all of (ia)-(v). In embodiments, the method comprises all of (ia)-(vi).

  In embodiments of any one of the foregoing methods, the method further includes identifying the level of a quality control marker, eg, in a second ROI, eg, a benign ROI. In embodiments, the quality control marker is selected from the tumor marker set, eg, DERL1.

  In embodiments, the method further comprises contacting the sample with a detection reagent for the quality control marker.

  In embodiments, the method further comprises, for example, directly outputting a signal that is related, eg, proportional, to the binding of the detection reagent to the first quality control marker, eg, in a second ROI, eg, a benign ROI. Or including obtaining indirectly.

  In embodiments, the method further includes identifying the level of a second quality control marker, eg, in a second ROI, eg, a benign ROI. In embodiments, the second quality control marker is other than a marker of the tumor marker set. In embodiments, the second quality control marker is associated with tumor lethality or aggressiveness. In embodiments, the second quality control marker is a marker described herein other than a marker of the tumor marker set, such as a tumor marker. In embodiments, the second quality control marker is selected from ACTN and VDAC1.

  In embodiments, the method further includes identifying the level of a third quality control marker, eg, in a second ROI, eg, a benign ROI. In embodiments, the third quality control marker is other than a marker of the tumor marker set. In embodiments, the third quality control marker is a marker described herein other than a marker of the tumor marker set, such as a tumor marker. In embodiments, the third quality control marker is selected from ACTN and VDAC1.

  In embodiments of any one of the foregoing methods, the method further comprises identifying a level, eg, amount, of a first quality control marker, eg, DERL1, in the second ROI, eg, a benign ROI; Identifying a second quality control marker within the two ROIs, eg, benign ROI, eg, the level of one of ACTN and VDAC.

  In embodiments, the method further includes identifying the level of one of a third quality control marker, eg, ACTN and VDAC, in the second ROI, eg, a benign ROI. In embodiments, the levels of the first, second and third quality control markers are identified in the same second ROI, eg, benign ROI. In embodiments, the levels of the first, second and third quality control markers are identified in different second ROIs, eg, different benign ROIs.

  In embodiments, the method further identifies the level of a first quality control marker, eg, DERL1, in a second ROI, eg, a benign ROI; a second in a second ROI, eg, a benign ROI. A quality control marker, eg, one level of ACTN and VDAC; and a third quality control marker, eg, one level of ACTN and VDAC, in a second ROI, eg, benign ROI Identifying, where, depending on the level, for example, classifying the sample as pass or not pass.

  In embodiments, the method includes detecting a level of signal of one of the quality control markers. In embodiments, the first value of the detected signal indicates a first quality level, eg, acceptable quality, and the second value of the detected signal indicates a second quality level, eg, unacceptable quality. In embodiments, depending on the value, the sample is processed or not processed, eg discarded, or the parameters of the analysis are changed.

  In embodiments of any one of the foregoing methods, the method comprises obtaining a multispectral image from the sample and the multispectral image in the following channel: a first ROI-specific detection reagent, eg, epithelial specific A marker channel; a second ROI-specific detection reagent, eg, a basal epithelial-specific marker channel; a nuclear-specific signal, eg, a DAPI signal channel; and a first population phenotype marker, eg, a first tumor marker Including unmixing in the channel. In embodiments, the method comprises: use of a first channel to collect a signal of a first ROI-specific detection reagent, eg, whole epithelial marker; a second ROI-specific detection reagent, eg, of a basal epithelial marker Use of a second channel to collect signals; use of a third channel to collect signals in the nuclear region; first population phenotype markers such as FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, Including the use of a fourth channel to collect the signal of a first tumor marker selected from CUL2 and HSPA9. In embodiments, the method further comprises: collecting a signal of a second population phenotypic marker, eg, a second tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9. Including the use of a fifth channel. In embodiments, the method further comprises: collecting a signal of a third population phenotypic marker, eg, a third tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9. Including the use of a sixth channel.

  In embodiments of any one of the foregoing methods, the method includes obtaining an image of the analyzed region of the sample, for example as a DAPI filtered image.

  In embodiments of any one of the foregoing methods, the method includes performing tissue localization, for example, by applying a tissue search algorithm to an image collected from the sample.

  In embodiments of any one of the foregoing methods, the method includes re-acquiring the image using DAPI and FITC monochrome filters.

  In embodiments of any one of the foregoing methods, the method comprises applying a tissue search algorithm to ensure that, for example, a pre-selected field-of-view image containing sufficient tissue is acquired. Including.

  In embodiments of any one of the foregoing methods, the method includes directly or indirectly obtaining continuous exposures of DAPI, FITC, TRITC, and Cy5 filters.

  In embodiments of any one of the foregoing methods, the method includes obtaining a multispectral image of the analyzed region of the sample.

  In embodiments of any one of the foregoing methods, the method comprises segmenting the sample region into epithelial cells, basal cells, and stroma.

  In embodiments of any one of the foregoing methods, the method further comprises identifying a cytoplasmic region and a nuclear region of the sample region.

  173. A method according to any one of the preceding claims comprising obtaining the value of a population phenotypic marker, such as a tumor marker, in the cytoplasm, nucleus and / or whole cell of a cancerous ROI, for example directly or indirectly .

  In embodiments of any one of the foregoing methods, the method comprises obtaining, for example, directly or indirectly the value of a population phenotypic marker, such as a tumor marker, in the cytoplasm, nucleus and / or whole cell of benign ROI. Including.

In embodiments of any one of the foregoing methods, the cancer or tumor sample comprises a plurality of portions, such as a plurality of sections or slices.

  In embodiments, the method comprises performing the steps described herein of a first population phenotypic marker, such as a first tumor marker, in a first portion, such as a section or slice, eg, collecting signals or Specification of acquiring or imaging, eg identifying levels; and second population phenotypic markers, eg second tumor markers, in a second portion, eg second section or slice Or collecting an image, or forming an image, eg, identifying a level. In embodiments, the second tumor marker is selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2, and HSPA9. In embodiments, the method further includes: identifying a ROI corresponding to a tumor epithelium in a second portion of the tumor sample, eg, a second section or slice; from the ROI corresponding to the tumor epithelium, FUS, For example, directly or indirectly obtaining a signal of a second tumor marker selected from SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9.

In embodiments, the method obtains a signal of an (ia) epithelial specific marker, eg, CK8, of said second portion of said tumor sample, eg, a second section or slice;
(Ii.a) obtaining a signal of a basal epithelial specific marker, eg CK5.

  In embodiments, the method further comprises: (ib) obtaining a signal of a second epithelial specific marker, eg, CK18, of the second portion of the tumor sample, eg, a second section or slice. (Ii.b) obtaining a signal of a second basal epithelial specific marker, such as TRIM29;

  In embodiments, the method further comprises obtaining a (iii) nuclear marker signal of the second portion of the tumor sample, eg, a second section or slice.

  In embodiments, the method further comprises obtaining a signal of the second tumor marker of claim 1; (iv) of the second portion of the tumor sample, eg, a second section or slice. Including. In embodiments, the method further comprises the second portion of the tumor sample, eg, a second section or slice; (v) the signal of the second tumor marker of the tumor marker set directly or indirectly. Including getting. In embodiments, the method further directly (directly) acquires a signal of a third tumor marker of the tumor marker set of (vi) the second portion of the tumor sample, eg, a second section or slice. Including doing. In embodiments, the method further comprises the second portion of the tumor sample, eg, the second section or slice; (vii) directly or indirectly the signal of the fourth tumor marker of the tumor marker set. Including getting. In embodiments, the method further comprises: directly or indirectly the signal of the fifth tumor marker of the tumor marker set of the second portion of the tumor sample, eg, a second section or slice; (viii) Including getting. In embodiments, the method further comprises the second portion of the tumor sample, eg, the second section or slice; (ix) the signal of the sixth tumor marker of the tumor marker set directly or indirectly. Including getting. In embodiments, the method further comprises the second portion of the tumor sample, eg, the second section or slice; (x) the signal of the seventh tumor marker of the tumor marker set directly or indirectly. Including getting. In embodiments, the method further comprises: directly or indirectly, the signal of the eighth tumor marker of the tumor marker set; or (xi) of the second portion of the tumor sample, eg, a second section or slice; Including getting.

  In embodiments, the signals of (ia) and (ib) have the same peak emission or are collected in the same channel.

In embodiments, the signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel.

  In embodiments, the method further comprises identifying an ROI corresponding to the tumor epithelium in a third portion of the tumor sample, eg, a third section or slice; from the ROI corresponding to the tumor epithelium, FUS, Including directly or indirectly obtaining a signal of a third tumor marker selected from SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9.

  In embodiments, the method comprises: (ii) obtaining a signal of an epithelial specific marker, eg, CK8, of a third portion of the tumor sample, eg, a third section or slice; ) Obtaining a signal of a basal epithelial specific marker, eg CK5 In embodiments, the method further comprises: (ib) obtaining a signal of a second epithelial specific marker, eg, CK18, of the third portion of the tumor sample, eg, a third section or slice. (Ii.b) obtaining a signal of a second basal epithelial specific marker, such as TRIM29; In embodiments, the method further comprises obtaining a signal of (iii) a nuclear marker of the third portion of the tumor sample, eg, a third section or slice. In embodiments, the method further comprises: (iv) obtaining a signal of the second tumor marker of claim 1 of the third portion of the tumor sample, eg, a third section or slice. Including. In embodiments, the signals of (ia) and (ib) have the same peak emission or are collected in the same channel. In embodiments, the signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel.

  In embodiments of any one of the foregoing methods, a portion of a first tumor sample, such as a first section or slice, is placed on a first substrate. In embodiments, a portion of a second tumor sample, such as a second section or slice, is placed on a second substrate. In embodiments, a portion of a third tumor sample, such as a third section or slice, is placed on a third substrate. In embodiments, a portion of a fourth tumor sample, such as a fourth section or slice, is placed on a fourth substrate.

  In embodiments, a portion of a first tumor sample, such as a first section or slice, and a portion of a second tumor sample, such as a second section or slice, are mounted on the same substrate.

  In embodiments of any one of the foregoing methods, the method further comprises digitally or electronically calculating a value corresponding to the signal, value, or image obtained from the sample at any step of the method described herein, digitally or electronically. Including storage or storage in a media database, for example in a computer.

  In embodiments of any one of the foregoing methods, the method exports a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, and Including identifying.

  In embodiments of any one of the foregoing methods, the method exports a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, to convert the cytoplasmic region into Including identifying.

  In embodiments of any one of the foregoing methods, the method exports values or images obtained by capturing signals from the tumor sample to software, eg, pattern recognition or object recognition software, to produce a cancerous ROI. Identifying.

  In embodiments of any one of the foregoing methods, the method exports a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, to generate a benign ROI. Including identifying.

  In embodiments of any one of the foregoing methods, the method exports values or images obtained by capturing signals from the tumor sample to software, eg, pattern recognition or object recognition software, to produce a cancerous ROI. Obtaining a value of the level of said tumor marker within.

  In embodiments of any one of the foregoing methods, the method exports a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, within a benign ROI. Obtaining a value for the level of said tumor marker.

  In embodiments of any one of the foregoing methods, the method comprises phenotypic markers of a region, such as a tumor marker signal, a first ROI marker, such as a whole epithelial specific marker signal and a second ROI marker, For example, obtaining a value for the level of a phenotypic marker, eg, a tumor marker, of a region within a cancerous ROI in response to a signal of a basal epithelial specific marker. In embodiments, the method includes calculating a risk score for the patient. In embodiments, the method includes calculating a risk score for the patient in response to the value.

  In embodiments of any one of the foregoing methods, the method comprises phenotypic markers of a region, such as a tumor marker signal, a first ROI marker, such as a whole epithelial specific marker signal and a second ROI marker, For example, obtaining a value for the level of a tumor marker within a benign ROI in response to a signal of a basal epithelial specific marker.

  In embodiments of any one of the foregoing methods, the method comprises phenotypic markers of a region, such as a tumor marker signal, a first ROI marker, such as a whole epithelial specific marker signal and a second ROI marker, For example, obtaining a cytoplasmic level value of a tumor marker in a cancerous ROI in response to a signal of a basal epithelial specific marker and a signal of a third ROI marker, eg, a nuclear specific marker.

  In embodiments of any one of the foregoing methods, the method comprises phenotypic markers of a region, such as a tumor marker signal, a first ROI marker, such as a whole epithelial specific marker signal and a second ROI marker, For example, obtaining a nuclear level value of a tumor marker in a benign ROI in response to a signal of a basal epithelial specific marker and a signal of a third ROI marker, eg, a nuclear specific marker.

  In embodiments of any one of the foregoing methods, the method includes calculating a risk score for the patient in response to one or more of the values.

  In embodiments, the method includes calculating a risk score for the patient, the risk score correlating with the likelihood of extraprostatic expansion or metastasis.

  In embodiments, the method includes performing a prognosis of the patient, classifying the patient, selecting a course of treatment for the patient, or selecting a selected course of treatment according to the risk score. Applying to the patient.

  In embodiments, the risk score corresponds to a “good prognosis” case (eg, surgical Gleason 3 + 3 or 3 with a minimum of 4, with organ-localized (≦ T2) tumor).

  In embodiments, the risk score corresponds to a “poor prognosis” case (eg, capsule penetration (T3a), seminal vesicle infiltration (T3b), lymph node metastasis, or a dominant Gleason pattern of 4 or higher).

  In embodiments, the risk score results in “good prognosis” cases (eg, surgical Gleason 3 + 3 or 3 with a minimum of 4 with organ-localized (≦ T2) tumor) and “poor prognosis” cases (eg, capsule penetration ( A distinction can be made between T3a), seminal vesicle invasion (T3b), lymph node metastasis or a dominant Gleason pattern of 4 or higher.

  In embodiments, the risk score is: surgical Gleason 3 + 3 or localized disease (≦ T3a) (defined as “low risk”); surgical Gleason ≧ 3 + 4 or nonlocalized disease (T3b, N or M ) (Defined as “medium to high risk”); surgical Gleason ≦ 3 + 4 and organ-localized disease (≦ T2) (defined as “good prognosis”); or surgical Gleason ≧ 4 + 3 or non-organ-localized disease (T3a, T3b, N or M) ("poor prognosis") or predicted.

  In embodiments where a risk score is calculated, the method further comprises identifying the patient as having an aggressive cancer or having an increased risk or cancer-related lethal outcome in response to the risk score. Including.

  In embodiments where a risk score is calculated, the method further comprises selecting an adjunctive therapy for the patient (eg, depending on the risk score) or giving the patient an adjunctive therapy.

  Also provided herein is a kit comprising 1, 2, 3, 4, 5, 6, 7 or all detection reagents of the tumor markers FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9 To do. In embodiments, the kit further comprises a detection reagent for whole epithelial markers and basal epithelial markers.

  Also herein, on a cancer sample, eg, a prostate tumor sample: a detection reagent for total epithelial marker; a detection reagent for basal epithelial marker; selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9 Provided is a cancer sample, eg, a prostate tumor sample, on which a detection marker for a tumor marker to be placed is placed. In embodiments, a cancer sample, eg, a prostate tumor sample, includes a plurality of portions, eg, slices.

  In embodiments, a detection reagent for a second tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9 is further disposed on a cancer sample, eg, a prostate tumor sample.

  The present invention also includes (i) identifying the ROI (cancerous ROI) of the tumor sample corresponding to the tumor epithelium; (ii) the following tumor markers in the cancerous ROI: FUS, SMAD4, DERL1, YBX1, pS6 , PDSS2, CUL2 and HSPA9 (tumor marker set) each level, eg, the amount, wherein the identification of the tumor marker level is related to, eg, proportional to the binding of an antibody to said tumor marker For example, directly or indirectly; obtaining a value for each level of the tumor marker within the cancerous ROI; and (iv) combining the levels, eg, by an algorithm Assigning a risk score to the patient, evaluating the tumor sample in response to the value, and And assessing prostate tumor samples by, and it is characterized in computer-implemented method of evaluation of the patient's prostate tumor samples.

  In embodiments, the method includes: using a first channel to collect signals for whole epithelial markers; using a second channel to collect signals for basal epithelial markers; collecting signals for nuclear regions Use of a third channel to collect a signal of a tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9.

  In embodiments, a first tumor marker level of the tumor marker set is identified in a first cancerous ROI, and a second tumor marker level of the tumor marker set is identified in a second cancerous ROI. The

In embodiments, both the level of the first tumor marker and the level of the second tumor marker from the tumor marker set are identified in the same cancerous ROI.

  In embodiments, the method further includes: identifying a level of a first quality control marker, eg, DERL1, in a second ROI, eg, a benign ROI; a second in a second ROI, eg, a benign ROI A quality control marker, eg, one level of ACTN and VDAC; and a third quality control marker, eg, one level of ACTN and VDAC, in a second ROI, eg, benign ROI Identifying, where, depending on the level, for example, classifying the sample as pass or not pass.

  The present invention provides a method for predicting the prognosis of cancer (eg, prostate cancer) in a patient (eg, a human patient). Confidence regarding whether the patient has an aggressive form of cancer or is at risk of having an aggressive form of cancer and / or whether the patient is at risk of having a cancer-related lethal outcome A realistic prediction is obtained.

  In some embodiments, the prognostic method of the invention is selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1, and YBX1 in a sample obtained from a patient. Measuring two or more prognostic determinants (PD) levels, wherein the level measurements indicate the prognosis of the cancer patient.

In some embodiments, the prognostic methods of the present invention in a sample taken from a patient:
(1) at least one cytoskeletal gene or protein (eg, α-actinin, such as α-actinin 1, 2, 3, and 4);
(2) at least one ubiquitinated gene or protein (eg, CUL1, CUL2, CUL3, CUL4A, CUL4B, CUL5, CUL7, DERL1, DERL2 and DERL3);
(3) at least one dependent receptor gene or protein (eg, DCC, neogenin, p75NTR, RET, TrkC, Ptc, EphA4, ALK and MET);
(4) at least one DNA repair gene or protein (eg, FUS, EWS, TAF15, SARF and TLS);
(5) at least one terpenoid skeleton biosynthesis gene or protein (eg, PDSS1 and PDSS2);
(6) at least one PI3K pathway gene or protein (eg, RpS6 and PLAG1);
(7) at least one TFG-β pathway gene or protein (eg, SMAD1, SMAD2, SMAD3, SMAD4, SMAD5 and SMAD9);
(8) at least one voltage-dependent anion channel gene or protein (eg, VDAC1, VDAC2, VDAC3, TOMM40 and TOMM40L); and / or (9) at least one RNA splicing gene or protein (eg, U2AF or YBX1) )
Measuring two or more PD levels selected from;
In this case, the level measurement indicates the prognosis of the cancer patient.

  The method may include the additional step of taking a sample (eg, a cancerous tissue sample) from the patient. The sample can be a solid tissue sample, such as a tumor sample. Solid tissue samples include formalin fixed paraffin embedded (FFPE) tissue samples, snap frozen tissue samples, ethanol fixed tissue samples, tissue samples fixed with organic solvents, tissue samples fixed with plastic or epoxy, cross-linked fixed tissue samples, surgery Tumor tissue removed by or a biopsy material sample, such as a core biopsy material, a resected tissue biopsy material, or an incision tissue biopsy material. In other embodiments, the sample may be a liquid sample, such as a blood sample and a circulating tumor cell (CTC) sample. In a further embodiment, the tissue sample is a prostate tissue sample, such as an FFPE prostate tumor sample.

  In some embodiments, in the prognostic method of the invention: at least ACTN1, YBX1, SMAD2 and FUS; at least ACTN1, YBX1 and SMAD2; at least ACTN1, YBX1 and FUS; at least ACTN1, SMAD2 and FUS; or at least YBX1, SMAD2 Measure RNA levels or protein levels of two or more PDs, including FUS.

  In some embodiments, the method of the invention measures at least 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 PDs. In a further embodiment, the method includes 3 types of PD (ie, PD1-3), 4 types of PD (ie, PD1-4), 5 types of PD (ie, PD1-5), 6 types of PD ( That is, PD1-6), 7 types of PD (ie, PD1-7), 8 types of PD (ie, PD1-8), 9 types of PD (ie, PD1-9), 10 types of PD (ie, PD1). PD1 to 10), 11 types of PD (ie, PD1 to 11) or 12 types of PD (ie, PD1 to 12) are measured, and at this time, the PDs are all different from each other, and are independently ACTN1, CUL2 , DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1.

  In some embodiments, in the prognostic method of the invention, one or more PDs whose levels are up-regulated relative to a reference value in aggressive forms of cancer or cancers at high risk of lethal outcome. Measure. Such PD can be, for example, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2, and VDAC1. In the method, one or more PDs whose levels are down-regulated relative to a reference value in aggressive forms of cancer or cancers at high risk of lethal outcome may be measured. Such PD can be, for example, ACTN1, RpS6, SMAD4 and YBX1.

  In a further embodiment, in the method of the present invention, in addition to PD selected from the aforementioned 12 types of biomarkers, HOXB13, FAK1, COX6C, FKBP5, PXN, AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, One or more PDs selected from the group consisting of EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5 and pPRAS40 are measured.

  The prognostic method of the present invention may measure the expression level of the selected PD, for example, using an antibody or an antigen-binding fragment thereof. Expression levels or protein levels can be measured by immunohistochemistry or immunofluorescence. For example, each antibody or antigen-binding fragment directed against PD can be labeled with a different fluorophore or conjugated with a different fluorophore, and the signals from the fluorophore can be separate or in parallel (multiplex) It can be detected by an automatic imaging device. In some embodiments, the tissue sample can be stained with DAPI. In some embodiments, the method can measure selected PD protein levels in the cellular component compartment, eg, in the nucleus, cytoplasm, or cell membrane. Alternatively, the measurement may be performed on whole cells.

  Measurements can be made in a defined region of interest of the tissue sample, eg, in a tumor region that excludes non-cancerous cells. For example, non-cancerous cells may be bound by binding to anti-cytokeratin 5 antibody and / or anti-TRIM29 antibody (eg, by staining) and / or lack of specific binding to anti-cytokeratin 8 antibody or anti-cytokeratin 18 antibody ( Not significantly higher than the background noise level). On the other hand, cancerous cells may be by binding to anti-cytokeratin 8 antibody and / or anti-cytokeratin 18 antibody (eg, by staining) and / or by lack of specific binding to anti-cytokeratin 5 antibody and anti-TRIM29 antibody. Can be identified. In one particular embodiment, the method comprises contacting a cross-section of an FFPE prostate tumor sample with an anti-cytokeratin 8 antibody, an anti-cytokeratin 18 antibody, an anti-cytokeratin 5 antibody, and an anti-TRIM29 antibody, wherein The measurement step is performed in a region in the cross section where the boundary is indicated by the anti-cytokeratin 8 antibody and the anti-cytokeratin 18 antibody, and the boundary is not indicated by the anti-cytokeratin 5 antibody and the anti-TRIM29 antibody.

  In some embodiments, in addition to measuring the biomarkers of the invention, at least one standard parameter associated with the subject cancer, such as Gleason score, tumor progression classification, tumor grade, tumor size, It may be desirable to evaluate tumor appearance characteristics, tumor location, tumor growth, lymph node status, tumor thickness (Breathrow score), ulceration, age of onset, PSA level and PSA kinetics obtain.

  The prognostic method of the present invention is clinically useful to improve the effectiveness of cancer treatment and to avoid unnecessary treatment. For example, the biomarkers and diagnostic methods of the present invention can be used to identify cancer patients in need of adjuvant therapy, taking a tissue sample from the patient; in the sample of the biomarkers described herein Measuring the level, and then patients with an aggressive cancer prognosis or patients with an increased risk of cancer-related lethal outcome may be treated with adjuvant therapy. Accordingly, the present invention also provides a method for treating cancer patients by identifying or selecting patients with poor prognosis when examined by the prognostic method of the present invention and treating only patients with poor prognosis with adjuvant therapy. Adjuvant therapy may be given to patients who have undergone standard care treatments such as surgery, radiation, chemotherapy or androgen removal. Examples of adjuvant therapy include, but are not limited to, radiation therapy, chemotherapy, immunotherapy, hormone therapy and targeted therapy. A targeted therapy may target one component of a signal transduction pathway in which one or more of the selected PDs are one component, where the targeted component is the selected PD and Are the same or different.

  Also, according to the present invention, to measure the level of two or more PDs selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1, and YBX1 A diagnostic kit comprising a reagent for specifically measuring the level of selected PD is provided. Reagents can include one or more antibodies or antigen-binding fragments thereof, oligonucleotides or aptamers. The reagent can be, for example, one that measures RNA transcript level or protein level of a selected PD.

  Also according to the present invention: (a) preparing a cell expressing PD selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1; b) contacting the cell with a candidate compound; and (c) examining whether the candidate compound modifies the expression or activity of the selected PD; wherein the observed is in the presence of the compound. Alterations identify compounds that can reduce the risk of cancer progression, reduce the risk of cancer progression, or delay or delay the progression of cancer, indicating that the compound can reduce the risk of cancer progression or delay or delay cancer progression Provide a way to do it. The compounds so identified can be used in the cancer treatment methods of the invention.

  The following embodiments are also described herein:

Embodiment 1. FIG. In a sample taken from a patient, two or more prognostic determinants (PD) selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1 Measuring the level of
In this case, the level measurement indicates the prognosis of the cancer patient,
A method for predicting the prognosis of a cancer patient.

Embodiment 2. FIG. In a sample taken from a patient, at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; At least one terpenoid skeleton biosynthesis gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion channel gene or protein; or at least one Measuring two or more PD levels selected from a class of RNA splicing genes or proteins;
In this case, the level measurement indicates the prognosis of the cancer patient,
A method for predicting the prognosis of a cancer patient.

Embodiment 3. FIG. The at least one cytoskeletal gene or protein is α-actinin 1, α-actinin 2, α-actinin 3 or α-actinin 4; the at least one ubiquitinated gene or protein is CUL1, CUL2, CUL3, CUL4A, CUL4B, CUL5, CUL7, DERL1, DERL2 or DERL3; the at least one dependent receptor gene or protein is DCC, neogenin, p75 ^NTR , RET, TrkC, Ptc, EphA4, ALK or MET; The at least one DNA repair gene or protein is FUS, EWS, TAF15, SARF or TLS; the at least one terpenoid skeleton biosynthesis gene or protein is PDSS1 or PDSS2. The at least one PI3K pathway gene or protein is RpS6 or PLAG1; the at least one TFG-β pathway gene or protein is SMAD1, SMAD2, SMAD3, SMAD4, SMAD5 or SMAD9; The method of embodiment 2, wherein said voltage-dependent anion channel gene or protein is VDAC1, VDAC2, VDAC3, TOMM40 or TOMM40L; or said at least one RNA splicing gene or protein is U2AF or YBX1.

  Embodiment 4 FIG. The method of any one of embodiments 1-3, further comprising the step of collecting a sample from the patient.

  Embodiment 5. FIG. The method of any one of embodiments 1-4, wherein the prognosis is that the cancer is an aggressive form of cancer.

  Embodiment 6. FIG. The method of any one of embodiments 1-4, wherein the prognosis is that the patient is at risk of having an aggressive form of cancer.

  Embodiment 7. FIG. The method of any one of embodiments 1-4, wherein the prognosis is that the patient is at risk of having a cancer-related lethal outcome.

Embodiment 8. FIG. Taking a tissue sample from the patient; and in said sample, ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1 or two selected from the group Including measuring the level of many PDs,
In this case, the level measurement indicates that the patient requires adjuvant therapy,
A method for identifying cancer patients in need of adjuvant therapy.

Embodiment 9. FIG. Collecting a tissue sample from a patient; and in which at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one At least one terpenoid skeleton biosynthesis gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion channel Measuring two or more PD levels selected from at least one RNA splicing gene or protein;
In this case, the level measurement indicates that the patient requires adjuvant therapy,
A method for identifying cancer patients in need of adjuvant therapy.

Embodiment 10 FIG. Measuring the level of two or more PDs selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1; and according to level measurements A method of treating a cancer patient comprising treating the patient with adjuvant therapy if the patient is shown to have an aggressive form of cancer or is at risk of having a cancer-related lethal outcome.

Embodiment 11. FIG. Identifying a patient having at least two PD level changes, wherein the level change is one or more of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2, and VDAC1 And a method of treating a cancer patient, comprising treating the patient with an adjuvant therapy; and

Embodiment 12 FIG. At least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; at least one terpenoid backbone biosynthesis From a gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-gated anion channel gene or protein; or from at least one RNA splicing gene or protein Measuring the level of two or more PDs selected from the group; and the level measurement causes the patient to have an aggressive form of cancer or a cancer-related condition If it has been shown that there is a risk of having a sexual outcome, including treating the patient with adjuvant therapy, a method of treating cancer patients.

  Embodiment 13 FIG. The method of any one of embodiments 8-12, wherein the adjuvant therapy is selected from the group consisting of radiation therapy, chemotherapy, immunotherapy, hormonal therapy and targeted therapy.

  Embodiment 14 FIG. Targeted therapy targets one component of the signal transduction pathway in which one or more of the selected PDs are one component, in which case the targeted component is the selected PD Different, the method of embodiment 13.

  Embodiment 15. FIG. 14. The method of embodiment 13, wherein the targeted therapy is one that targets one or more of the selected PDs.

  Embodiment 16. FIG. The method of any one of embodiments 8-12, wherein the patient has been subjected to standard care treatment.

  Embodiment 17. FIG. Embodiment 17. The method of embodiment 16 wherein the standard care treatment is surgery, radiation, chemotherapy or androgen removal.

  Embodiment 18. FIG. The method of any one of embodiments 1-17, wherein the patient is a patient with prostate cancer.

Embodiment 19. FIG. The two or more PDs are:
A) at least ACTN1, YBX1, SMAD2 and FUS;
B) at least ACTN1, YBX1 and SMAD2;
C) at least ACTN1, YBX1 and FUS;
D) at least ACTN1, SMAD2 and FUS; or E) at least YBX1, SMAD2 and FUS
including,
The method of any one of Embodiments 1-18.

  Embodiment 20. FIG. The method of any one of embodiments 1-19, wherein at least 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 PDs are selected.

  Embodiment 21. FIG. Six types of PDs consisting of PD1, PD2, PD3, PD4, PD5 and PD6 are selected, and these PD1, PD2, PD3, PD4, PD5 and PD6 are different and are independently ACTN1, CUL2, DCC, DERL1 The method of any one of embodiments 1, 4-8, 9, 10, and 12-19, selected from the group consisting of: FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1, and YBX1.

  Embodiment 22. FIG. Seven types of PD consisting of PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are selected, and PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are different, and are independently ACTN1, CUL2 Embodiments 1, 4-8, 9, 10 and 12-19 selected from the group consisting of: DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1.

  Embodiment 23. FIG. Further, the group consisting of HOXB13, FAK1, COX6C, FKBP5, PXN, AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5 and pPRAS40, or pPRAS40. The method of any one of embodiments 1, 5-8, 10 and 13-22, comprising measuring the level of more PD.

  Embodiment 24. FIG. 24. The method as in any one of embodiments 1-23, wherein at least one level measurement of the selected two or more PDs is up-regulated relative to a reference value.

  Embodiment 25. FIG. 25. The method of any one of embodiments 1-24, wherein at least one level measurement of the selected two or more PDs is down-regulated relative to a reference value.

  Embodiment 26. FIG. At least one level measurement of the selected two or more PDs is up-regulated relative to a reference value, and at least one of the selected two or more PDs 26. The method of any one of embodiments 1-25, wherein one is down-regulated relative to a reference value.

  Embodiment 27. FIG. 25. The method of embodiment 24, wherein the selected PD includes one or more PDs selected from the group consisting of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2, and VDAC1.

  Embodiment 28. FIG. 26. The method of embodiment 25, wherein the selected PDs include one or more PDs selected from the group consisting of ACTN1, RpS6, SMAD4, and YBX1.

  Embodiment 29. FIG. The method of any one of embodiments 1-28, wherein the measuring step comprises measuring the protein level of the selected PD.

  Embodiment 30. FIG. 30. The method of embodiment 29, wherein the protein level is measured by an antibody or fragment thereof.

  Embodiment 31. FIG. The method of embodiment 30, wherein the protein level is measured by immunohistochemistry or immunofluorescence.

  Embodiment 32. FIG. The method of embodiment 30, wherein the antibody or fragment thereof is each labeled with a different fluorophore or conjugated to a different fluorophore and the signal from the fluorophore is detected in parallel by an automated imaging device.

  Embodiment 33. FIG. The method of embodiment 32, wherein the tissue sample is stained with DAPI.

  Embodiment 34. FIG. 30. The method of embodiment 29, wherein the measuring step comprises measuring the protein level of the selected PD within the cellular component compartment.

  Embodiment 35. FIG. 30. The method of embodiment 29, wherein the measuring step comprises measuring the protein level of the selected PD in the nucleus, in the cytoplasm or on the cell membrane.

  Embodiment 36. FIG. The method of any one of the preceding embodiments, wherein the level of PD is measured in a defined region of interest.

  Embodiment 37. FIG. The method of embodiment 36, wherein non-cancerous cells are excluded from the region of interest.

  Embodiment 38. FIG. 38. The method of embodiment 37, wherein non-cancerous cells are bordered by anti-cytokeratin 5 antibody and anti-TRIM29 antibody.

  Embodiment 39. FIG. 39. The method of embodiment 38, wherein the non-cancerous cells are not bounded by anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody.

  Embodiment 40. FIG. 40. The method of any one of embodiments 36-39, wherein cancerous cells are included in the region of interest.

  Embodiment 41. FIG. 44. The method of embodiment 43, wherein the cancerous cells are bordered by anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody.

  Embodiment 42. FIG. 42. The method of embodiment 41, wherein the cancerous cells are not bounded by anti-cytokeratin 5 antibody and anti-TRIM29 antibody.

  Embodiment 43. FIG. The method of any one of the preceding embodiments, wherein the measuring step comprises separately measuring the level of the selected PD.

  Embodiment 44. FIG. The method of any one of the preceding embodiments, wherein said measuring step comprises measuring the level of selected PD in a multiplex reaction.

  Embodiment 45. FIG. The method of any one of the preceding embodiments, wherein the sample is a solid tissue sample.

  Embodiment 46. FIG. Solid tissue samples are removed by formalin fixed paraffin embedded tissue samples, snap frozen tissue samples, ethanol fixed tissue samples, tissue samples fixed with organic solvents, tissue samples fixed with plastic or epoxy, cross-linked fixed tissue samples, surgery The method of embodiment 45, wherein the method is a tumor tissue or biopsy sample.

  Embodiment 47. 47. The method of embodiment 46, wherein the biopsy sample is a core biopsy material, a resected tissue biopsy material, or an incision tissue biopsy material.

  Embodiment 48. FIG. The method of any one of the preceding embodiments, wherein the tissue sample is a cancerous tissue sample.

  Embodiment 49. The method of any one of the preceding embodiments, wherein the tissue sample is a prostate tissue sample.

  Embodiment 50. FIG. 50. The method of embodiment 49, wherein the prostate tissue sample is a formalin fixed paraffin embedded (FFPE) prostate tumor sample.

  Embodiment 51. FIG. Furthermore, contacting the cross-section of the FFPE prostate tumor sample with an anti-cytokeratin 8 antibody, an anti-cytokeratin 18 antibody, an anti-cytokeratin 5 antibody and an anti-TRIM29 antibody, wherein the measuring step comprises: The method of embodiment 50, wherein the method is performed in a region demarcated by anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody and not delimited by anti-cytokeratin 5 antibody and anti-TRIM29 antibody.

  Embodiment 52. FIG. The method of any one of the preceding embodiments, further comprising measuring at least one standard parameter associated with the cancer.

  Embodiment 53. FIG. The at least one standard parameter includes Gleason score, tumor progression classification, tumor grade, tumor size, tumor appearance characteristics, tumor location, tumor growth, lymph node status, tumor thickness (Breathrow score) ), Ulceration, age of onset, PSA level and PSA kinetics.

  Embodiment 54. FIG. A kit for measuring the level of two or more PDs selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1, A kit comprising a reagent for specifically measuring the level of selected PD.

  Embodiment 55. FIG. 55. The kit of embodiment 54, wherein the reagent comprises one or more antibodies or fragments thereof, oligonucleotides or aptamers.

  Embodiment 56. FIG. 56. The kit of embodiment 54, wherein the reagent measures RNA transcript level or protein level of a selected PD.

Embodiment 57. FIG. (A) preparing a cell expressing PD selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1;
(B) contacting the cell with a candidate compound; and (c) examining whether the candidate compound modifies the expression or activity of the selected PD;
Here, the modifications observed in the presence of the compound indicate that the compound can reduce the risk of cancer progression or delay or delay cancer progression.
A method of identifying a compound that can reduce the risk of cancer progression, or that can delay or retard cancer progression.

Embodiment 58. FIG. Measuring the level of PD selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1; and an agent that modulates the level of the selected PD A method for treating cancer patients, comprising administering

Embodiment 59. At least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; at least one terpenoid backbone biosynthesis From a gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-gated anion channel gene or protein; or from at least one RNA splicing gene or protein Measuring the level of two or more PDs selected from the group; and administering an agent that modulates the level of the selected PD Method of treating a cancer patient.

Embodiment 60. FIG. Identifying a patient having at least two PD level changes, wherein the level change is one or more of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2, and VDAC1 Selected from the group consisting of an upregulation of and a downregulation of one or more of ACTN1, RpS6, SMAD4 and YBX1; and administering an agent that modulates the level of at least one of the PDs A method for treating a cancer patient, comprising:

  Embodiment 61. FIG. A method for defining a region of interest in a tissue sample comprising contacting the tissue sample with one or more first reagents for specifically identifying the region of interest.

  Embodiment 62. FIG. 62. The method of embodiment 61, wherein the region of interest comprises cancerous cells.

  Embodiment 63. FIG. 63. The method of embodiment 62, wherein the one or more first reagents comprise an anti-cytokeratin 8 antibody and an anti-cytokeratin 18 antibody.

  Embodiment 64. FIG. Further, defining a tissue sample region to be excluded from the region of interest by contacting the tissue sample with one or more second reagents to specifically identify the region to be excluded. The method of any one of Embodiments 61-63.

  Embodiment 65. The method of embodiment 64, wherein the area to be excluded comprises non-cancerous cells.

  Embodiment 66. FIG. The method of embodiment 65, wherein said one or more second reagents comprise an anti-cytokeratin 5 antibody and an anti-TRIM29 antibody.

  Aspects and embodiments also relate to computer-implemented or automated methods of evaluating tumor samples, for example, for assigning risk scores to patients.

  Aspects and embodiments also relate to a system that includes a memory and a processing device functional in evaluating a tumor sample, eg, for assigning a risk score to a patient.

  Aspects and embodiments also relate to a system that includes a memory and a processing device functional for evaluating a tumor sample, for example, to analyze signals from a complete tumor sample or to assign a risk score to a patient.

  Also, aspects and embodiments, when executed on a computer processor, implement a method for evaluating a tumor sample, for example, to analyze a signal from a complete tumor sample or to assign a risk score to a patient The present invention relates to a computer-readable medium containing computer-executable instructions.

  US Provisional Patent Application No. 61 / 792,003, to which this application claims priority, includes at least one drawing made in color. Copies of US Provisional Application No. 61 / 792,003 containing color drawing (s) will be provided by the US Patent Office upon request and payment of the necessary fee.

FIG. 1 shows a hematoxylin and eosin stained section of a surgically removed prostate tumor. An American Board of Pathology certified anatomical pathologist annotated the sections and identified four regions with the highest measured Gleason score pattern and two regions with the lowest measured value. One of the high measurement cores is extracted from the tumor sample for incorporation into the high measurement tissue microarray (TMA), and one of the low measurement cores is extracted from the tumor sample for incorporation into the low measurement value TMA did.

FIG. 2 shows a biomarker selection and validation engine that can be used to identify biomarkers for any disease or condition. The engine has three phases: a biological phase, a technical phase and a performance phase. MoAb-monoclonal antibody; DAB-3,3'-diaminobenzidine; IF-immunofluorescence; and TMA-tissue microarray.

FIG. 3 shows the prostate cancer specific biomarker selection and validation engine. The engine has three phases: a biological phase, a technical phase and a performance phase. Initially, 160 potential biomarkers were identified. Using this biomarker selection and validation engine, 12 markers were identified as being correlated with tumor invasion. MoAb-monoclonal antibody; DAB-3,3'-diaminobenzidine; IF-immunofluorescence; and TMA-tissue microarray.

FIG. 4 shows the inter-section reproducibility using quantitative multiplex immunofluorescence in the control cell line TMA (CTMA). CTMA sections 27 and 41 were stained with immunofluorescent antibodies against FUS-N and DERL1. The fluorescence intensity of each cell line in CTMA was compared between sections 27 and 41 and the results were graphed as indicated. The linear relationship of the amount of immunofluorescence in the two cell lines and the high R ² value indicate the reproducibility of the quantitative immunofluorescence assay between experiments.

FIG. 5 shows the details of the cohort of samples included in the low measurement TMA with respect to tumor invasion and lethal outcome. Of the 297 patients included in the tumor invasion trial, 110 patients have indolent tumors, 122 patients have moderate tumors, and 67 patients based on surgical Gleason scores Had an aggressive tumor. Of the 317 patients included in the lethal outcome study, 275 patients had indolent tumors (not died of prostate cancer) and 42 patients had aggressive tumors (prostate cancer or Died of distant metastasis). The first five columns show clinical data, while the last four columns give an estimate of the number of samples that would be useful if the model was trained with 3, 6, 9 or 12 markers.

FIG. 6 shows the inter-system reproducibility of the two Vectra Intelligent Slide Analysis Systems. CTMA was evaluated for Alexa-568, Alexa-633, and Alexa-647 detection in duplicate in two different systems. The two systems differed by about 7% for Alexa-568 detection, about 20% for Alexa-633 detection and about 2% for Alexa-647 detection. Anti-VDAC1, FUS and SMAD4 antibodies were used for Alexa-568, 633 and 647 channels, respectively.

FIG. 7 shows automatic image acquisition, processing by the Vectra Intelligent Slide Analysis System, and automatic image analysis by Definience Developer XD ^™ . The biomarker intensity score obtained by automated analysis can then be used to examine biomarker correlations by bioinformatics or to evaluate clinical samples using the Harvest Laboratory Information System (LIS).

FIG. 8 shows quality control features incorporated into automatic image analysis, where each image is analyzed from each phosphor layer to detect oversaturation, abnormal texture or lack of tissue. The region labeled “Artifact” indicates detection of supersaturation such that the supersaturated region is excluded from image analysis.

FIGS. 9A-9F show automatic identification of a region of interest (ROI) using the Definiens Developer XD ^™ . FIG. 9A shows a raw image imported into a system containing multiple channels of fluorescence. FIG. 9B shows that tumor epithelial structures are identified based on anti-cytokeratin 8 and anti-cytokeratin 18 staining. FIG. 9C shows that overlaying the nucleus identifies where cells are localized within the tumor epithelial region. FIG. 9D shows that the cells are defined as benign or malignant based on the presence of the basal cell markers cytokeratin 5 and TRIM29. FIG. 9E shows that areas of benign and malignant tumors are defined. FIG. 9F shows that the region of interest is defined based on the location of benign and malignant tumors.

FIG. 10 shows the quantification of biomarker (PD) immunofluorescence within the region of interest. Note that two biomarkers (DERL1 (PD1) and FUS (PD2)) are expressed at lower levels in the malignant tumor region than in the benign tumor region.

FIG. 11 shows 17 biomarkers that showed univariate performance for predicting tumor invasion and lethal outcome in the HLTMA trial with a low core. The core Gleason score is the measured Gleason score. The most reliable results were obtained when cores from moderate tumors (based on surgical Gleason score) were excluded. By defining cores from moderate tumors as indolent or aggressive, the correlation between biomarkers and tumor invasion can be distorted towards indolent or aggressive associations.

FIG. 12 shows bioinformatics analysis of HLTMA test data.

FIG. 13 shows the frequency of which biomarkers appear in the top 1% of combinations sorted by AIC for correlation with tumor invasion. The frequencies are shown for combinations of up to 3 biomarkers, up to 4 biomarkers, up to 5 biomarkers, up to 6 biomarkers, up to 8 biomarkers and up to 10 biomarkers. . The biomarkers tested were selected from a pool of 17 biomarkers pre-selected for univariate performance in the mini TMA assay and HLTMA.

FIG. 14 shows the frequency of which biomarkers appear in the top 5% of combinations sorted by AIC for correlation with tumor invasion. The frequencies are shown for combinations of up to 3 biomarkers, up to 4 biomarkers, up to 5 biomarkers, up to 6 biomarkers, up to 8 biomarkers and up to 10 biomarkers. . The biomarkers tested were selected from a pool of 17 biomarkers pre-selected for univariate performance in the mini TMA assay and HLTMA.

FIG. 15 shows the frequency of which biomarkers appear in the top 1% and top 5% of up to 7 member combinations sorted by AIC and test data for correlation with tumor invasion. The biomarkers tested were selected from a pool of 17 biomarkers that were pre-selected for univariate performance in the mini TMA assay with HLTMA and HLTMA.

FIG. 16 shows the frequency of which biomarkers appear in the top 1% of up to 5 member combinations sorted by AIC and test data for correlation with tumor invasion. The biomarkers tested were selected from a pool of 31 biomarkers that were not pre-selected for univariate performance in HLTMA.

FIG. 17 shows the frequency of which biomarkers appear in the top 5% of up to 5 member combinations sorted by AIC and test data for correlation with tumor invasion. The biomarkers tested were selected from a pool of 31 biomarkers that were not pre-selected for univariate performance in HLTMA.

FIG. 18 shows the top 12 types of markers for each type of analysis and the concordance between the top markers for various analyses. Seven biomarker cores were identified as appearing in the top 12 marker lists at 75% or 100% of the analysis. We also identified a secondary set of 7 biomarkers as appearing in the top 12 marker lists in 50% of the analyses.

FIG. 19 shows the frequency of which biomarkers appear in the top 1% of combinations sorted by AIC for correlation with lethal outcome. The frequencies are shown for combinations of up to 3 biomarkers, up to 4 biomarkers, up to 5 biomarkers, up to 6 biomarkers, up to 8 biomarkers and up to 10 biomarkers. . The biomarkers tested were selected from a pool of 17 biomarkers that were preselected for univariate performance in HLTMA.

FIG. 20 shows the frequency of which biomarkers appear in the top 5% of combinations sorted by AIC for correlation with lethal outcome. The frequencies are shown for combinations of up to 3 biomarkers, up to 4 biomarkers, up to 5 biomarkers, up to 6 biomarkers, up to 8 biomarkers and up to 10 biomarkers. . The biomarkers tested were selected from a pool of 17 biomarkers that were preselected for univariate performance in HLTMA.

FIG. 21 shows the frequency of which biomarkers appear in the top 1% and top 5% of up to 7 member combinations sorted by AIC and test data for correlation with lethal outcome. The biomarkers tested were selected from a pool of 17 biomarkers that were preselected for univariate performance in HLTMA.

FIG. 22 shows that markers that partially overlap in correlation with tumor invasion and lethal outcome can potentially be used to evaluate both endpoints in a single assay. For example, as shown in FIGS. 11, 13 and 19, ACTN1 and YBX1 show a high degree of correlation with both tumor invasion and lethal outcome.

FIG. 23 shows a triplex analysis that can be used to evaluate three biomarkers (PD) in addition to tumor mask markers and nuclear staining on a single slide. The first biomarker PD1 can be detected using a FITC-conjugated primary antibody and an anti-FITC-Alexa 568 secondary antibody. The second biomarker PD2 can be detected using a rabbit primary antibody, a biotin-conjugated anti-rabbit secondary antibody and streptavidin conjugated to Alexa 633. The third biomarker PD3 can be detected using mouse primary antibody, horseradish peroxidase (HRP) conjugated anti-mouse secondary antibody and anti-HRP-Alexa 647 tertiary antibody. For tumor masks, anti-CK8-Alexa 488 and anti-CK18-Alexa 488 can be used to identify tumor epithelial structures, and anti-CK5-Alexa 555 and anti-TRIM29-Alexa 555 are used to identify basal cell markers. Can be done. Tumor section quality can be assessed by general autofluorescence (AFL) and autofluorescence from red blood cells and bright granules (BAFL). Any three types of biomarkers (PD) can be used for triple staining (provided the correct antibody combination is available), but in this figure, PD1 is HSD17B4 and PD2 is FUS PD3 is LATS2.

FIG. 24 shows biomarker antibody combinations that can be combined for triple analysis. Using such a combination, twelve biomarkers can be evaluated in four sections from a tumor sample.

FIG. 25 shows that the observed interference is minimal when antibodies for multiple biomarkers (SMAD (PD1) and RpS6 (PD2)) are used in the same assay. The linear relationship between the amount of immunofluorescence in the two assays and the high R ² value indicate that the interference by the second antibody to the first is minimal.

FIG. 26 illustrates an exemplary computer system in which various aspects of embodiments of the invention can be implemented.

FIGS. 27A-F outline an automated quantitative multiplex immunofluorescence experimental approach and biomarker measurements within a defined region of interest in prostatectomy tissue. FIGS. 27A-F outline an automated quantitative multiplex immunofluorescence experimental approach and biomarker measurements within a defined region of interest in prostatectomy tissue. FIGS. 27A-F outline an automated quantitative multiplex immunofluorescence experimental approach and biomarker measurements within a defined region of interest in prostatectomy tissue.

  FIG. 27A shows the spectral profile of each fluorophore in the spectral library used in the assay and the profile of tissue autofluorescence (AFL) and bright autofluorescence (BAFL) signals, respectively.

  FIG. 27B shows a general outline of a staining procedure for quantitative multiplex immunofluorescence biomarker measurements in a tissue region of interest. SPP1 and SMAD4 were used as examples. Region of interest marker antibodies (CK8 and CK18 for whole epithelium and CK5 and TRIM29 for basal epithelium) were conjugated directly to Alexa488 and Alexa555. Biomarker antibodies were detected using the secondary and tertiary antibody sequences described. The color in the table indicates the position of the special spectrum of the emission peak of the displayed Alexa fluorphore dye.

  FIG. 27C shows that composite multispectral image (i) is unmixed into individual channels corresponding to autofluorescence (AFL) and bright autofluorescence (BAFL), region of interest markers and biomarkers as shown (ii). Indicates.

  FIG. 27D shows segmentation and biomarker quantification of Definiens script line tissue. When parts 1 to 6 are moved from the composite image (1), the entire epithelial region is first identified (2), followed by the nuclear region (3). The epithelial area was further segmented into tumors (which were visualized in red), benign (which was visualized in green), and obscure (which was visualized in yellow) (4). Gray color showed non-epithelial areas such as stroma and blood vessels (4). Finally, biomarkers were quantified from the tumor epithelial area only (which was a red outline) (5 and 6).

  FIG. 27E shows an organization annotation and quality control procedure. Left: Representative hematoxylin of a human prostatectomy sample showing four (blue) and two (green) 1 mm diameter circles placed on the highest and lowest Gleason pattern areas, respectively, annotated by a professional pathologist And eosin (H & E) stained sections. Two cores (each with a diameter of 1 mm) were collected from four blue regions to produce a TMA block. Right: Serial sections of the same prostatectomy sample were stained with DAPI and CK8 / CK18-Alexa488. Areas with bright staining of prostate epithelium with CK8 / 18 cytokeratin antibody are considered good quality areas, while areas with little or no staining (as indicated by yellow spotted areas) are considered low quality and are suitable for TMA construction Not considered.

FIG. 27F shows in-experiment reproducibility. Two serial sections from the prostate tumor test TMA were stained in the same experiment. Images were acquired using a Vectra system and processed using Definiens script. Compare the mean values of staining intensity of CK8 / 18, PTEN and SMAD4 in the same core of serial TMA sections in a scatter plot. Linear regression curves, equations and R ² values were generated using Excel software.

Figures 28A-C show a covariate description and univariate analysis of lethal outcome. FIG. 28A shows the composition of the lethal outcome-annotated prostatectomy cohort used in this study and a comparison with the DHS et al., Nature 2011, 470: 269-273 PHS cohort. FIG. 28B shows Kaplan-Meier curves of survival as a function of single biomarker protein expression in the test cohort. The population with the highest third of the risk score was divided from the population with the lower two third risk score. Annotate P value (P) and hazard ratio (HR).

Figures 29A-C show the development of multivariate models and Kaplan-Meier survival plots. FIG. 29A shows multivariate Cox regression and logistic regression analysis of survival prediction in this study cohort. Using a combination of markers, models were developed based on training and testing in the entire cohort. Four types of markers: PTEN, SMAD4, CCND1, SPP 1.3 types of markers: SMAD4, CCND1, SPP1,. FIG. 29B shows Kaplan-Meier curves of survival as a function of risk score generated by the Cox model trained with 4 markers or [3 markers + pS6 + pPRAS40] in all cohorts. A minimum of two-thirds of the risk score was used as the threshold for population division. FIG. 29C shows a lethal outcome-predictive performance comparison of the four markers (PTEN, SMAD4, CCND1, SPP1) between this study and Ding et al.

Figures 30A-E show validation of PTEN, CCND1, SMAD4, SPP1, P-S6 and P-PRAS40 antibody specificities. Doxycycline inducible shRNA knockdown cell lines were established for PTEN (FIG. 30A), CCND1 (FIG. 30B) and SMAD4 (FIG. 30C). Doxycycline treatment reduced abundant target protein as assessed by Western blotting (WB) in all cases. Cell lines with high or low / negative level expression of of PTEN (Figure 30A), CCND1 (Figure 30B) and SMAD4 (Figure 30C) were also examined by WB and immunohistochemistry (IHC) to determine antibody specificity. Further verification. SPP1 (FIG. 30D) antibody detected an SPP1-specific band and an additional band of low molecular weight (as assessed by WB in PC3 cells), while the upper SPP1-specific band was significantly reduced in low SPP1-expressing BxPC3 cells. The intensity of SPP1 antibody staining in PC3 and BxPC3 cells by IHC correlated well with the relative intensity of the SPP1-specific band detected by WB. The specificity of P-S6 and P-PRAS40 antibodies (FIG. 30E) was verified in DU145 cells. LY294002 treatment significantly reduced phosphorylation of S6 and PRAS40 as shown by WB and IHC, respectively.

FIG. 31 shows an outline of a statistical analysis flow. For each patient, two tissue cores from the region with the highest Gleason score were placed in the TMA block. The average value of biomarker expression within the tumor epithelial region of each TMA core was used for analysis, and two biomarker values were obtained per patient. For PTEN, SMAD4 and pS6, the lowest values in the two cores were used for the analysis. For CCND1, SPP1, p90RSK, pPRAS40 and Foxo3a, the highest values from the two cores were used. Using such values, 10,000 bootstrap training samples were created and both multivariate Cox and logistic regression models were trained on each training sample. Each test was performed in a complement set. Given the inclusion of censored data in the cohort, we estimated model performance using both the Concordance Index (CI) and “Area Under the Curve” (AUC). The marker combinations tested in the model were as follows: 4 types of markers (PTEN, SMAD4, CCND1, SPP1), 3 types of markers (SMAD4, CCND1, SPP1), and 3 types of markers and the following phosphomarkers : Each combination of pS6, pPRAS40 and [pS6 + pPRAS40].

FIG. 32 shows the creation of a biopsy simulated tissue microarray (TMA). Tissue blocks from prostatectomy samples were annotated with all visible Gleason patterns (top). The example shown is from a patient with a global Gleason score (GS) of 4 + 3 = 7. As shown in the higher magnification figure (center), the patterns within the same block can be very diverse. Two 1 mm cores were taken from each tissue block. One was taken from the region with the highest GS (4 + 4 = 8) and embedded in agarose / paraffin with high scoring cores from other blocks to create HTMA (bottom left). The other was taken from the region with the lowest GS (3 + 3 = 6) and embedded in agarose / paraffin with low scoring cores from other blocks to create LTMA (bottom right).

FIG. 33 shows a biomarker selection strategy. Three types of criteria (biological, technical and performance based) were used to select 12 final biomarkers. (DAB: Ab specificity evaluated based on chromogenic tissue staining with diaminobenzidine (DAB); IF: Ab specificity and performance based on immunofluorescent tissue staining).

FIGS. 34A and 34B show univariates of 39 biomarkers for disease aggressiveness and disease-specific mortality measured in both low (L TMA; black bar) and high (H TMA; brown bar) Gleason regions. Show performance. FIG. 34A shows the odds ratio (OR) for predicting severe disease pathology (aggressiveness) calculated for each marker. Markers with OR on the left side of the vertical line are negatively correlated with disease severity as assessed by disease state. The right side of this line is positively correlated. Markers were ranked based on OR when measured in LTMA. FIG. 34B shows hazard ratios for disease death (lethal) calculated for each marker and plotted as described in FIG. 34A. In FIGS. 34A and 34B, biomarkers with two asterisks ( ^** ) show statistical significance at the 0.1 level in both LTMA and HTMA. Biomarkers with a single asterisk ( ^* ) show statistical significance in HTMA only, and LTMA has no significance. Note the large overlap between biomarkers and statistically significant univariate performance in both aggressive disease and death from disease.

FIG. 35A and FIG. 35B show a performance-based biomarker selection process for disease aggressiveness. FIG. 35A shows that the bioinformatics workflow has selected the most frequently used biomarkers from all combinations of up to 5 markers from 31 sets. FIG. 35B shows an example of the performance of a top-rank 5-marker model, including comparison with training in L TMA and then testing in independent samples from L TMA and H TMA. Note that test performance in LTMA and HTMA is consistent with substantial overlap of confidence intervals. FIG. 35C shows that combinations were created that allow for up to 3, 4 or 5 biomarkers. This figure shows the most frequently included proteins when predicting aggressive disease ranked by test using a 5 biomarker model. FIG. 35A and FIG. 35B show a performance-based biomarker selection process for disease aggressiveness. FIG. 35A shows that the bioinformatics workflow has selected the most frequently used biomarkers from all combinations of up to 5 markers from 31 sets. FIG. 35B shows an example of the performance of a top-rank 5-marker model, including comparison with training in L TMA and then testing in independent samples from L TMA and H TMA. Note that test performance in LTMA and HTMA is consistent with substantial overlap of confidence intervals. FIG. 35C shows that combinations were created that allow for up to 3, 4 or 5 biomarkers. This figure shows the most frequently included proteins when predicting aggressive disease ranked by test using a 5 biomarker model.

36A and 36B show the final biomarker set and selection criteria. FIG. 36A shows 12 biotypes selected based on univariate performance on aggressiveness (indicated by OR on the left) and lethality and frequency of occurrence in multivariate models for disease aggressiveness or lethal outcome (right side of the table). Marker is shown. FIG. 36B summarizes the biomarker names and biological significance. This set of biomarkers includes proteins known to function in the regulation of cell proliferation, cell survival and metabolism. FIG. 36C shows that a multivariate 12 marker model for disease aggressiveness was developed based on logistic regression. The obtained AUC and OR are shown. Subsequently, the risk score generated by the aggressive model for all patients was correlated with the lethal outcome. The obtained AUC and HR are shown. 36A and 36B show the final biomarker set and selection criteria. FIG. 36A shows 12 biotypes selected based on univariate performance on aggressiveness (indicated by OR on the left) and lethality and frequency of occurrence in multivariate models for disease aggressiveness or lethal outcome (right side of the table). Marker is shown. FIG. 36B summarizes the biomarker names and biological significance. This set of biomarkers includes proteins known to function in the regulation of cell proliferation, cell survival and metabolism. FIG. 36C shows that a multivariate 12 marker model for disease aggressiveness was developed based on logistic regression. The obtained AUC and OR are shown. Subsequently, the risk score generated by the aggressive model for all patients was correlated with the lethal outcome. The obtained AUC and HR are shown.

Figures 37A-L show the specificity of the antibodies. ACTN1 (FIG. 37A), CUL2 (FIG. 37B), Delrin 1 (FIG. 37C), FUS (FIG. 37D), PDSS2 (FIG. 37E), SMAD2 (FIG. 37F), VDAC1 (FIG. 37G) and YBX1 (FIG. 37H) antibodies Specificity was verified by Western blotting (WB) and immunohistochemistry (IHC) of siRNA treated and control cells. Marker-specific siRNA treatment significantly reduced the intensity of the band on WB, and the specificity of the antibody was confirmed by intracellular specific IHC staining. The specificity of the SMAD4 antibody (FIG. 37I) was verified by WB and IHC of SMAD4 positive control cell line PC3 and SMAD4 negative control cell line BxPC3. The specificity of the pS6 antibody (FIG. 37J) was verified by WB and IHC of untreated DU145 cells and DU145 cells treated with the PI3K inhibitor LY294002. LY294002 treatment significantly reduced phosphorylation of S6 as shown by WB and IHC. A band was detected on WB by the Leica anti-DCC antibody (FIG. 37K), which did not match the size predicted for the DCC protein (K designation “X”); and IHC staining was observed in DCC siRNA treated cells Not reduced (left panel in FIG. 37K). The Leica anti-DCC antibody appeared to recognize the HSPA9 protein as shown by WB and IHC of HSPA9 siRNA treated and control cells (right panel of FIG. 37K). β-actin was used as a WB loading control.

Figures 38A-G show the identification of HSPA9 but not DCC as a prognostic biomarker for prostate cancer. Leica anti-DCC antibodies were not verified by DCC siRNA knockdown cells by WB and IHC (FIG. 38A). That was because the size of the band detected by the antibody on WB was much smaller than expected for the DCC protein (75 kDa versus 158 kDa) and IHC staining intensity was not reduced in DCC siRNA treated cells. . By mass spectrometry, the protein recognized by Leica anti-DCC antibody on WB was identified as HSPA9. To confirm that the Leica anti-DCC antibody was indeed an anti-HSPA9 antibody, it was tested by WB and IHC in HSPA9 siRNA treated cells and control cells; which was the IHC staining detected by the WB band and Leica anti-DCC antibody? Was also significantly reduced in HSPA9 siRNA treated cells (FIG. 38B). The WB and IHC patterns of Leica anti-DCC antibodies in HSPA9 siRNA treated cells were similar to those detected by Santa Cruz anti-HSPA9 antibody (FIG. 38C). Silencing of HSPA9 by siRNA appears to reduce HeLa cell proliferation (FIG. 38D), reduced HeLa cell colony formation in clonogenic assays (FIG. 38E), cell death (FIG. 38F) and caspase activity (FIG. 38D). Caused an increase of 38G).

Figures 39A-C show the model design for the 8-marker indicator assay. FIG. 39A shows the odds ratio (with 95% confidence interval) for individual biomarkers. Quantitative biomarker measurements correlated with prostate cancer pathology as an endpoint. Note that the effective size is normalized. FIG. 39B shows the biomarker usage frequency in the top 10% of the multivariate model. Given that many models have similar performance in the bootstrap test AUC, the frequency of presence in the comprehensive top marker model search is used as an additional criterion for selecting the final marker for the diagnostic test. This figure shows how often the top 8 markers are present in the top 10% of the 8 marker model when sorted by the bootstrap test AUC median. FIG. 39C shows the final marker model coefficients shown on a continuous scale between 0 and 1 used in the logistic regression model for risk score calculation. Note that a negative sign indicates a protection marker. The unit of this coefficient is the fluorescence intensity scale associated with the assay.

  FIGS. 40A-F show clinical validation tests and performance for predicting favorable prognosis. Sensitivity and specificity curves (FIGS. 40A and 40B, respectively) can be used to identify appropriate risk taxa. The risk score distribution relative to the NCCN risk taxon (FIGS. 40C and 40D) and the D′ Amico risk taxon (FIGS. 40E and 40F) is significantly greater within each NCCN or D′ Amico level by the biomarker indicator assay. Indicates that risk information is added.

FIG. 40A shows that the relationship between sensitivity and associated medical decision level can be used to identify low risk taxa. For example, a good prognosis classification can be identified as a patient with a risk score between 0 and 0.33, which is 90% (95% CI, 82% to 94%) sensitive (P [risk score> 0. 33 | pathology with poor prognosis]). In this case, patients with poor prognosis may have a chance of 10% (95% CI, 6% to 18%) accidentally receiving a good prognosis classification. This false negative can lead to inadequate treatment.

FIG. 40B shows that the relationship between specificity and the level of medical decision associated with can be used to identify poor prognostic taxa as well. For example, a poor prognosis classification can be identified as a patient with a risk score in the interval (0.8-1), with an interval of 95% (95% CI, 90% -98%) specificity (P [risk score ≦ 0.80 | good disease state])). In this case, patients with good prognosis may have a chance of 5% (95% CI, 2% to 10%) accidentally receiving a poor prognosis classification. This false positive can lead to overtreatment.

FIG. 40C is included in the risk score cutoff level with a median risk score of 0.33-0.8 derived using each NCCN risk level (very low, low, medium, high) biomarker indicator assay. Shows that the predictive value (+ PV) for a good prognosis (surgical Gleason 3 + 3 or 3 + 4 and ≦ T2) disease state was found at 85% with a risk score cutoff <0.33. The predictive value (−PV) for a poor prognosis condition was 100% with a risk score cutoff> 0.9 and 76.9% with a risk score> 0.8. At a risk score <0.33, 95% of patients with a “very low” NCCN classification had a good prognosis, but the frequency measurement of good prognosis cases with a “very low” NCCN classification alone was 80.3% Met. In the “low” NCCN category, at a risk score <0.33, 81.5% of patients had good prognosis, but the frequency measurement of good prognosis according to the “low” NCCN criteria was 63.8%. It was. Conversely, at a risk score> 0.8, 75% of patients in the “very low” NCCN category have a poor prognosis, and when the risk score is> 0.8, 76.9% of all patients He had a poor prognosis.

FIG. 40D shows frequency measurements of good prognosis cases as a function of risk score quartiles. The increase in risk score quartile was highly correlated with the decrease in good prognosis frequency observed in each NCCN category. Furthermore, frequency measurements for patients with good prognosis identified by test vs. NCCN stratification alone increased from 0% to 23.8% at an 81% confidence level.

FIG. 40E was included in the risk score cutoff level with a median risk score of 0.33-0.8 derived using the biomarker indicator assay at each D'Amico risk level (low, medium, high) It shows that. The predictive value (+ PV) for a favorable prognosis is 85% with a risk score cutoff <0.33. The predictive value (−PV) for poor prognosis is 100% with a risk score cutoff> 0.9, and 76.9% with a risk score> 0.8. At a risk score <0.33, 87.2% of patients with the “low” D′ Amico classification have a good prognosis, while the frequency measurement of good prognosis cases in the “low” D′ Amico group is 70. 6%. In the “intermediate” D′ Amico category, at a risk score <0.33, 75% of patients have a good prognosis but frequency measurements of all patients with a good prognosis in the “intermediate” D′ Amico group Is 41.2%. Conversely, at a risk score> 0.8, 59.3% of patients in the “low” D′ Amico category have a poor prognosis, and when the risk score is> 0.8, 76. 9% have a poor prognosis.

FIG. 40F shows frequency measurements of good prognosis cases as a function of risk score quartiles. Increased risk score quartiles were highly correlated with decreased prognostic case frequency measurements in each D'Amico category. Furthermore, frequency measurements for patients with good prognosis identified by test vs. D'Amico stratification alone increased from 0% to 23.8% with a confidence level of 81%.

FIGS. 41A-D show the results of performance results for the clinical validation study, the entire cohort: “GS6” pathology (surgical Gleason = 3 + 3 and local ≦ T3a, N = 256). FIG. 41A shows test sensitivity (P [risk score> threshold | “non-GS6” pathology]) as a function of medical decision level. FIG. 41B shows the risk score specificity (P [risk score <threshold || GS6 ”pathology]) to identify the“ non-GS6 ”category. 41C and 41D show the distribution of risk scores for “GS6” and “non-GS6” pathologies. FIG. 41E shows the receiver operating characteristic (ROC) curve of this model. Area under the ROC curve (AUC) = 0.65 (95% confidence interval [CI], 0.58-0.72), P <0.001, and highest to lowest quartile odds ratio (OR) = 4. 2 (95% CI, 1.9-9.3). The OR of the quantitative risk score was 12.59 (95% CI, 3.5-47.2) / unit change.

FIGS. 42A-C show performance results for prediction of clinical validation study, whole cohort: good prognosis (surgical Gleason ≦ 3 + 4 and organ localization ≦ T2, N = 274). FIG. 42A shows the distribution of risk scores for conditions with good prognosis. FIG. 42B shows the distribution of risk scores for poor prognosis conditions. FIG. 42C shows the ROC curve of this model. AUC = 0.68 (95% CI, 0.61-0.74), P <0.0001, and highest to lowest quartile OR = 3.3 (95% CI, 1.8-6.1) . The OR of the quantitative risk score was 20.9 (95% CI, 6.4-68.2) / unit change.

FIGS. 43A-C are cohorts with clinical validation trials, National Comprehensive Cancer Center Network (NCCN) and D'Amico criteria: performance for good prognosis (surgical Gleason ≦ 3 + 4 and organ-localized ≦ T2, N = 256) The results are shown. FIG. 43A shows the distribution of risk scores for disease with good prognosis. FIG. 43B shows the distribution of risk scores for poor prognosis diseases. FIG. 43C shows the ROC curve of this model. AUC = 0.69 (95% CI, 0.63-0.73), P <0.0001, and highest to lowest quartile OR = 5.5 (95% CI, 2.5-12. 1) . The OR of the quantitative risk score was 26.2 (95% CI, 7.6-90.1) / unit change.

44A-B show how the net reclassification improvement analysis shows the good index (risk score ≦ 0.33) and poor prognosis (risk score> 0.8) molecular index categories NCCN (FIG. 44A) and D′ Amico. FIG. 44B shows showing that the SOC risk classification system can be supplemented. Patients with a molecular risk score ≦ 0.33 in the NCCN low, medium and high or D′ Amico medium and high categories may be considered to be at lower risk for aggressive disease than those shown in the SOC category alone. Patients with a molecular risk score> 0.8 in NCCN very low, low and intermediate and D'Amico low and intermediate categories may be considered higher risk of aggressive disease than shown in the SOC category alone. A molecular risk score ≦ 0.33 for the categories NCCN very low and D′ Amico low can be considered as confirmatory. Similarly, a molecular risk score> 0.8 for categories NCCN high and D'Amico high can be considered as confirmatory. Note that patients with good prognosis in the left square and poor prognosis in the right square reflect the correct risk adjustment. Of patients with favorable prognosis, 78% and 53% of NCCN and D'Amico, respectively, are correctly adjusted. Of patients with poor prognosis, 76% and 88% of NCCN and D'Amico, respectively, are correctly adjusted. Also note that patients with a molecular risk score ≤0.33 and very low category NCCN and low D'Amico have significantly more prognostic patients than the overall risk group. R. S. = Molecular risk score. 44A-B show how the net reclassification improvement analysis shows the good index (risk score ≦ 0.33) and poor prognosis (risk score> 0.8) molecular index categories NCCN (FIG. 44A) and D′ Amico. FIG. 44B shows showing that the SOC risk classification system can be supplemented. Patients with a molecular risk score ≦ 0.33 in the NCCN low, medium and high or D′ Amico medium and high categories may be considered to be at lower risk for aggressive disease than those shown in the SOC category alone. Patients with a molecular risk score> 0.8 in NCCN very low, low and intermediate and D'Amico low and intermediate categories may be considered higher risk of aggressive disease than shown in the SOC category alone. A molecular risk score ≦ 0.33 for the categories NCCN very low and D′ Amico low can be considered as confirmatory. Similarly, a molecular risk score> 0.8 for categories NCCN high and D'Amico high can be considered as confirmatory. Note that patients with good prognosis in the left square and poor prognosis in the right square reflect the correct risk adjustment. Of patients with favorable prognosis, 78% and 53% of NCCN and D'Amico, respectively, are correctly adjusted. Of patients with poor prognosis, 76% and 88% of NCCN and D'Amico, respectively, are correctly adjusted. Also note that patients with a molecular risk score ≤0.33 and very low category NCCN and low D'Amico have significantly more prognostic patients than the overall risk group. R. S. = Molecular risk score.

FIG. 45A shows an overview of all four quantitative multiplex immunofluorescent triplex assay formats (PBXA / B / C / D) for staining of 12 markers. The region of interest marker antibody was conjugated directly with Alexa dye, while the biomarker antibody in channel 568 was conjugated with fluorescein isothiocyanate (FITC). All biomarkers (primary antibodies) were detected with secondary and tertiary antibody sequences other than pS6 and PDSS2 (which were conjugated directly to FITC). Each color corresponds to a specific channel. Biomarkers with an asterisk ( ^* ) were used for internal tissue quality control purposes, and those with ACTN1, DERL1 or VDAC below a given signal intensity were automatically excluded. The eight biomarkers that are used in the prediction algorithm for quantitative measurement in the tumor epithelium are shown in italics. FIG. 45B shows that during the image acquisition process, an image of the entire slide is first acquired with a mosaic of 4 × monochrome 4 ′, 6-diamidino-2-phenylindole (DAPI) filtered images. Tissue search algorithm is used to determine the position of the tissue, in which case the image is reacquired using both 4xDAPI and 4xFITC monochrome filters, and later using another tissue search algorithm, everything is sufficient A 20 × field image containing a significant amount of tissue was acquired using continuous exposure of DAPI, FITC, tetramethylrhodamine isothiocyanate (TRITC) and Cy5 filters. Image cubes were stored for automatic unmixing into individual channels and further processing by Definiens software. FIG. 45C shows the various steps of the total quantitative multiplex immunofluorescence assay procedure. The raw slide was first visually inspected for staining and the presence of dye with a fluorescence microscope. The presence of a recognizable amount of fluorescent dye excluded the slide from further analysis. Tissues that passed the initial quality control were subjected to a multiplex staining procedure followed by image acquisition, Definiens analysis and bioinformatics analysis. The image acquisition process was performed as described above for FIG. 45B. Image cubes were stored in the server, unmixed into individual channels, and processed by Definiens software. Data were collected by Definiens software using ROI biomarkers from the tumor and benign regions of each specific region of interest (ROI). If the bioinformatics analysis algorithm was below the predetermined signal intensity for ACTN1, DERL1 or VDAC 1, it was excluded before further analysis of the data.

DETAILED DESCRIPTION OF THE INVENTION The present invention is from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1 ("prognostic determinants" or "PD"; Table 1). A biomarker panel comprising two or more members selected is based on the finding that it is useful in obtaining a reliable prognosis of molecular evidence systems for cancer patients. By measuring expression in a cancerous tissue sample from a patient (eg, protein expression) or the level of biomarker activity, the tumor is aggressive in the cancer patient, eg, the ability of the tumor to infiltrate the surrounding tissue or the risk of progression Can be predicted with confidence. Cancer progression is indicated by, for example, cancer metastasis or recurrence. The level also predicts the lethal outcome of cancer or the effectiveness of cancer treatment (eg, surgery, radiation therapy or chemotherapy) independently of or in addition to traditional established risk assessment procedures Can also be used to Levels can also be used to identify patients in need of aggressive cancer treatment (eg, chemotherapy given in addition to surgical treatment) or to guide further diagnostic tests. When used in the context of a pathway status gene or protein, the level also provides patients with information about the type of treatment that may be most beneficial to cancer patients and for inclusion in clinical studies Can also be used to stratify. Levels can also be used to identify patients who are not and / or do not need cancer treatment (eg, surgery, radiation therapy, chemotherapy, targeted therapy or adjuvant therapy). In other words, the biomarker panel of the present invention enables a doctor to optimally manage cancer patients.

  In some embodiments, the primary clinical indication of a multiplex or multivariate diagnostic method of the invention accurately predicts whether PCA is “aggressive” (eg, active prostate tumors at diagnosis). By predicting the probability of progression (ie, “active aggressive disease”; or progression at some later time (ie, “progression risk”)) or “slow” or “dormant” Another clinical indication of the method may be to accurately predict the probability that a patient will die from PCA (ie, “lethal outcome” / “disease-specific death”). In a model that assigns a risk score to a sample, the C statistic is the ratio of the number of sample pairs of one aggressive sample and one indolent sample, where the aggressive sample is , Such samples Have a higher risk scores than indolent samples in the total number of pairs.

Definitions “Acquiring” or “obtaining”, as the term is used herein, refers to ownership of a physical entity (eg, sample) or value, eg, a numerical value or image, It is obtained by “obtaining” or “obtaining indirectly”. “Direct acquisition” means performing a process (eg, performing a synthesis or analysis method, contacting a sample with a detection reagent, or capturing a signal from a sample) to obtain a physical entity or value. To do. “Indirect acquisition” refers to receiving a physical entity or value from another party or source (eg, from a third party laboratory that directly acquired the physical entity or value). Acquiring a physical entity directly includes performing a process that includes physical changes in a physical material. Exemplary changes include creating a physical entity from two or more starting materials, shearing or fragmenting the material, separating or purifying the material, two or more separate Or a chemical reaction that includes noncovalent bonds or breakage or formation of noncovalent bonds. Obtaining a value directly includes performing a process that includes a physical change in a sample or another substance, for example, an analytical process that includes a physical change in a substance, such as a sample, analyte, or reagent (sometimes described herein An analysis method, for example: separation or purification of a substance, eg a test substance or a fragment or other derivative thereof, from another substance; test substance or a fragment thereof Or combining another derivative with another substance, such as a buffer, solvent, or reactant; or the structure of the analyte or fragment or other derivative thereof, for example, the first atom and the second of the analyte By breaking or forming a covalent or non-covalent bond between atoms; inducing a signal, eg a light system signal, eg a fluorescent signal or Or by altering the structure of the reagent or fragments or other derivatives thereof (eg, by breaking or forming a covalent or non-covalent bond between the first atom and the second atom of the reagent) ) Performing a method that includes one or more of the changing. Acquiring values directly includes methods in which a computer or detection device, such as a scanner, is used when there is a change in electronic state in response to, for example, a photon impact on a detector. Acquiring values directly involves capturing the signal from the sample.

  A detection reagent, as used herein, has sufficient specificity for its intended target and can be used to distinguish the target from others discussed herein Typically a binding reagent. In embodiments, the detection reagent is one that has no or substantially no binding to other (non-target) species under the conditions of performing the method.

  A region of interest (ROI), as the term is used herein, is one or more entities such as a single cell entity (eg, a cell component component (eg, nucleus or cytoplasm), tissue component, single cell connective tissue matrix , A single-cell colloid substance, an extracellular component, such as interstitial tissue fluid), or a substance thereof containing a region-phenotype marker, wherein the region-phenotype marker is ROI or a sample, tissue or patient from which the same was obtained It refers to the cell used for analysis. In one embodiment, the ROI entity is a cell.

  A region-phenotype marker, as the term is used herein, is one that reflects, predicts or is associated with a preselected phenotype, eg, cancer, eg, a cancer subtype or patient outcome. In one embodiment, the region-phenotype marker is one that reflects, predicts or is associated with an inflammatory disorder (eg, an autoimmune disorder), a nervous system disorder or an infectious disease. In one embodiment, the pre-selected phenotype is one that exists or affects an ROI entity or cell. In one embodiment, the pre-selected phenotype is a ROI, sample, tissue phenotype of the target disease for which the patient is analyzed, eg, cancer.

  By way of example, ROIs include cancer cells, such as cancerous prostate cells, the preselected phenotype is that of cancerous cells, and the population-phenotype marker is a cancer marker, such as prostate cancer, It is a tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9.

  As used herein, unless indicated otherwise in the text, pS6 refers to the phosphorylated form of ribosomal protein S6 and is encoded by the RpS6 gene.

  In one embodiment, the first ROI is a cancerous ROI and the second ROI is a benign ROI.

  In one embodiment, the region-phenotype marker is expressed in ROI cells. In one embodiment, the region-phenotype marker is present in a cell of ROI but is not expressed in the cell, eg, in one embodiment, the region-phenotype marker is a secretory factor found in the stroma. Yes, and therefore the stroma is ROI in this example.

The ROI can be obtained in various ways. As an example, the ROI is:
A first tissue type or cell type having a pre-selected relationship with a morphological feature, eg, a second tissue type or cell type, eg, surrounded by a second tissue type or cell type;
For example, by the possession of a selected molecule, such as a protein, mRNA or DNA (referred to herein as a ROI marker) marker, by non-morphological characteristics, eg by possessing molecular characteristics; or by morphological characteristics and non-morphological It can be selected or identified by a combination of features.

  In one embodiment, identification or selection by morphological characteristics includes selection of ROI from other cells or substances, eg, other tissues, eg, ROI from non-cancerous cells, eg, by incision of a cancerous region ( For example, by manual or automated means) and physical separation. In one embodiment of morphological selection, such as microdissection, the ROI is removed from the environment in an essentially intact state. In one embodiment of morphological selection, such as microdissection, the ROI is removed, but the morphological structure is not maintained.

  In one embodiment of selection or identification by non-morphological features, an ROI can be identified or selected by including, for example, ROI markers, eg, preselected molecular species, that are related in the ROI entity, eg, cells. As an example, cell sorting, eg FACS, can be used to obtain an ROI with non-morphological features. In one embodiment, FACS is used to separate ROI marker-bearing cells from other cells to obtain ROI.

  In one embodiment of ROI selection by a combination of morphological and non-morphological selection, a morphologically identifiable structure that exhibits a preselected binding pattern for a detection reagent for the ROI marker is an ROI. Used to get.

  An ROI includes an entity, typically a cell, from which a population phenotypic marker performs its function. In one embodiment, the ROI is a collection of entities, typically a collection of cells, from which a signal related to, eg, proportional to, a region-phenotype marker can be extracted. Samples can be evaluated by the level of the region-phenotype marker in the ROI. For example, in the case of prostate cancer, the level of one of the region-phenotype markers, eg, tumor markers, eg, tumor markers described herein, allows for the evaluation of the sample and the patient from whom the sample was taken. In one embodiment, the region-phenotype marker is selected based on exerting a function, eg, a function associated with the disorder being evaluated, and is prognostic or diagnosed in an ROI entity or cell. ROI markers are used in some embodiments to select or define an ROI.

  In one embodiment, the ROI is a collection of entities, typically a collection of cells, that form a pattern, for example in a patient (although not necessarily in a sample), eg, a different morphological region.

  A sample, as the term is used herein, is a composition comprising patient-derived cellular components or single-cell components. The term sample encompasses a raw sample from a patient, such as a biopsy sample, a processed sample, such as an immobilized tissue, a tissue-derived fraction or other material. The ROI is considered a sample.

Prognostic determinants In a first aspect of the invention, there is provided for use in determining cancer treatment. The terms “prognostic determinant”, “biomarker” and “marker” are used interchangeably herein and are analytes (eg, peptides or proteins) that can be objectively measured and evaluated as indicators of biological processes. ). The inventors have found that the level of expression or activity of such biomarkers is reliably correlated with the prognosis of cancer patients, such as tumor aggressive or lethal outcome. The possibility that such biomarkers can be correlated with cancer prognosis can be amplified by using them in combination.

  The at least one biomarker can be a cytoskeletal gene or protein. Without being bound by theory, cytoskeletal genes and proteins correlate with cancer prognosis because malignant tumors are characterized in part by tumor invasion into adjacent tissues and tumor spread to distant tumors. obtain. Such invasion and spreading typically requires reorganization of the cytoskeleton. Non-limiting examples of cytoskeletal genes and proteins useful as cancer prognostic biomarkers include α-actin, β-actin, γ-actin, α-actinin 1, α-actinin 2, α-actinin 3, α- Actinin 4, vinculin, E-cadherin, vimentin, keratin 1, keratin 2, keratin 3, keratin 4, keratin 5, keratin 6, keratin 7, keratin 8, keratin 9, keratin 10, keratin 11, keratin 12, keratin 13, Keratin 14, keratin 15, keratin 16, keratin 17, keratin 18, keratin 19, keratin 20, lamin A, lamin B1, lambin B2, lamin C, α-tubulin, β-tubulin, γ-tubulin , Δ-tubulin, ε-tubulin, LMO7, LATS And LATS2 and the like. Preferably, the cytoskeletal gene or protein is α-actinin 1, α-actinin 2, α-actinin 3 or α-actinin 4, especially α-actinin 1. α-actinin 1 has been shown to interact with CDK5R1; CDK5R2; collagen, type XVII, α1; GIPC1; PDLIM1; protein kinase N1; SSX2IP; and zyxin. These genes and proteins are therefore considered as cytoskeletal proteins for the interpretation of this application.

  The at least one biomarker can be a ubiquitinated gene or protein. Without being bound by theory, ubiquitinated genes and proteins can correlate with the prognosis of cancer because ubiquitin can bind to the protein and cause the protein to be destroyed towards the proteasome. Increasing the rate of protein synthesis often requires supporting transformation events in cancer, and protein regulatory processes such as ubiquitination are crucial in tumor progression. Non-limiting examples of ubiquitinated genes and proteins useful as cancer prognostic biomarkers include ubiquitin activating enzymes (eg, UBA1, UBA2, UBA3, UBA5, UBA6, UBA7, ATG7, NAE1 and SAE1), ubiquitin bound enzyme (e.g., UBE2A, UBE2B, UBE2C, UBE2D1, UBE2D2, UBE2D3, UBE2E1, UBE2E2, UBE2E3, UBE2G1, UBE2G2, UBE2H, UBE2I, UBE2J1, UBE2L3, UBE2L6, UBE2M, UBE2N, UBE2O, UBE2R2, UBE2V1, UBE2V2 and BIRC6 ), Ubiquitin ligase (for example, UBE3A, UBE3B, UBE3C, UBE4A, UBE4B, UBOX5, UBR5) WWP1, WWP2, mdm2, APC, UBR5, SOCS, CBLL1, HERC1, HERC2, HUWE1, NEDD4, NEDD4L, PPIL2, PRPF19, PIAS1, PIAS2, PIAS3, PIAS4, RANBPSM, RUS1 , F-box proteins (eg, cdc4), Skp1, cullin family members (eg, CUL1, CUL2, CUL3, CUL4A, CUL4B, CUL5, CUL7 and ANAPC2), RING proteins (eg, RBX1), elongin C and Endoplasmic reticulum-related degradation protein (“ERAD”, eg, DERL1, DERL2, DERL3, Doa10, EDE , ER mannosidase I, VIMP, SEL1, HRD1 and HERP) and the like. Preferably, the ubiquitinated gene or protein is a culin, in particular a CUL2 or ERAD gene or protein, in particular DERL1. CUL2 has been shown to interact with DCUN1D1, SAP130, CAND1, RBX1, TCEB2, and von Hippel-Lindau tumor suppressors. These genes and proteins are therefore considered ubiquitinated proteins for the interpretation of this application.

The at least one biomarker can be a dependent receptor gene or protein. Without being bound by theory, dependent receptor genes and proteins correlate with cancer prognosis because of two opposite signaling pathways: 1) cell survival, migration and differentiation; and 2) its ability to induce apoptosis. Can do. In the presence of a ligand, this receptor activates classical signaling pathways involved in cell survival, migration and differentiation. In the absence of ligand, it does not remain inactive; it instead induces an apoptotic signal. Cell survival, migration and apoptosis are all involved in cancer. Non-limiting examples of useful dependent receptor genes and proteins as biomarkers for the prognosis of cancer, DCC, ^{neogenin, p75 NTR, RET, TrkC,} Ptc, EphA4, ALK, it includes MET and integrin subsets. Preferably, the dependent receptor gene or protein is DCC. DCC has been shown to interact with PTK2, APPL1, MAZ, caspase-3, NTN1 and the androgen receptor. Accordingly, these genes and proteins are considered interpretive dependent receptor proteins for the present application.

  The at least one biomarker can be a DNA repair gene or protein. Without being bound by theory, DNA repair genes and proteins correlate with cancer prognosis because cells that have accumulated a large amount of DNA damage or cells that no longer cause damage to DNA can no longer be repaired effectively can enter unregulated cell division. Can do. Non-limiting examples of DNA repair genes and proteins useful as cancer prognostic biomarkers include homologous recombination repair genes and proteins (eg, BRCA1, BRCA2, ATM, MRE11, BLM, WRN, RECQ4, FANCA, FANCB FANCC, FANCD1, FANCD2, FANCE, FANCF, FANCG, FANCI, FANJ, FANCL, FANCM and FANCN), nucleotide excision repair genes and proteins (eg, XPC, XPE (DDB2), XPA, XPB, XPD, XPF and XPG) Non-homologous end joining genes and proteins (eg, NBS, Rad50, DNA-PKcs, Ku70 and Ku80), cross-injury synthetic genes and proteins (eg, XPV POLH)), mismatch repair genes and proteins (eg hMSH2, hMSH6, hMLH1, hPMS2), adenine base excision repair genes and proteins (eg MUTYH), cell cycle checkpoint genes and proteins (eg p53, p21, ATM, ATR, BRCA1, MDC1 and 53BP1) and TET family genes and proteins (eg, FUS, EWS, TAF15, SARF and TLS). Preferably, the DNA repair gene or protein is a TET family member, especially FUS. FUS has been shown to interact with FUSIP1, ILF3, PRMT1, RELA, SPI1 and TNPO1. Accordingly, these genes and proteins are considered as DNA repair proteins for the purposes of this application.

The at least one biomarker can be a terpenoid skeleton biosynthesis gene or protein. Without being bound by theory, terpenoid skeleton biosynthetic genes and proteins can be correlated with cancer prognosis, as reportedly, the biosynthesis of some terpenoids, such as CoQ ₁₀ , is reduced in cancer. Non-limiting examples of terpenoid skeleton biosynthetic genes and proteins useful as cancer prognostic biomarkers include ACAT1, ACAT2, HMGCS1, HMGCS2, HMGCR, MVK, PMMVK, MVD, IDI1, IDI2, FDPS, GGPS1, PDSS1 , PDSS2, DHDDS, FNTA, FNTB, RCE1, ZMPSTE24, ICMT, and PCYOX1. Preferably, the terpenoid skeleton biosynthetic gene or protein is PDSS2.

  The at least one biomarker can be a phosphatidylinostide 3-kinase (PI3K) pathway gene or protein. Without being bound by theory, PI3K genes and proteins can correlate with cancer prognosis because the pathway regulates apoptosis in part. Non-limiting examples of the PI3K pathway include ligands (eg, insulin, IGF-1, IGF-2, EGF, PDGF, FGF and VEGF), receptor tyrosine kinases (eg, insulin receptor, IGF receptor, EGF). Receptors, PDGF receptors, FGF receptors and VEGF receptors), kinases (eg PI3K, AKT, mTOR, GSK3-β, IKK, PDK1, CDKN1B, FAK1 and S6K), phosphatases (eg PTEN and PHLPP), Ribosomal proteins (eg, ribosomal protein S6), adapter proteins (eg, GAB2, GRB2, GRAP, GRAP2, PIK3AP1, PRAS40, PXN, SHB, SH2B1, SH2B2, SH2B3, SH2D3A and SH D3C) immunophilins (e.g., FKBP12, FKBP52 and FKBP5) and transcription factors (e.g., FoxO1, Hif1-alpha, DEC1 and PLAG1) and the like. Preferably, the PI3K gene or protein is a ribosomal protein such as ribosomal protein S6, especially phospho-rpS6, or a transcription factor gene or protein, especially PLAG1. PLAG1 has been shown to regulate transcription of IGF-2 and other target genes such as CRLF1, CRABP2, CRIP2 and PIGF. Thus, CRLF1, CRABP2, CRIP2 and PIGF are considered PI3K proteins for the interpretation of this application.

  The at least one biomarker can be a transforming growth factor-β (TGF-β) pathway gene or protein. Without being bound by theory, TGF-β genes and proteins can correlate with cancer prognosis because the TGF-β signaling pathway stops the cell cycle at the G1 phase to stop proliferation and promote apoptosis. Disruption of TGF-β signaling increases proliferation and reduces apoptosis. Non-limiting examples of members of the TGF-β pathway include ligands (eg, activin A, GDF1, GDF11, BMP2, BMP3, BMP4, BMP5, BMP6, BMP7, Nodal, TGF-β1, TGF-β2 and TGF- β3), type I receptors (eg TGF-βR1, ACVR1B, ACVR1C, BMPR1A and BMPR1B), type II receptors (eg TGF-βR2, ACVR2A, ACVR2B, BMPR2B), SARA, receptor-regulated SMAD (eg SMAD1, SMAD2, SMAD3, SMAD5 and SMAD9), coSMAD (eg SMAD4), apoptotic proteins (eg DAXX) and cell cycle proteins (eg p15, p21, Rb and c-myc). . Preferably, the TGF-β pathway gene or protein is SMAD, in particular SMAD2 or SMAD4.

  The at least one biomarker can be a voltage-gated anion channel gene or protein. Without being bound by theory, voltage-gated anion channel genes and proteins have been shown to contribute to apoptosis and can therefore correlate with cancer prognosis. Non-limiting examples of voltage-dependent anion channels include VDAC1, VDAC2, VDAC3, TOMM40, and TOMM40L. Preferably, the voltage-dependent anion channel is VDAC1. VDAC1 has been shown to interact with gelsolin, BCL2-like 1, PRKCE, Bcl-2 binding X protein and DYNLT3. These genes and proteins are therefore considered as voltage-gated anion channels for the interpretation of this application.

  The at least one biomarker can be an RNA splicing gene or protein. Without being bound by theory, RNA splicing genes and proteins can correlate with cancer prognosis because aberrant splicing mRNA is also found in a high proportion of cancerous cells. Non-limiting examples of RNA splicing genes and proteins include snRNP (eg, U1, U2, U4, U5, U6, U11, U12, U4atac and U6atac), U2AF and YBX1. Preferably, the RNA splicing gene or protein is YBX1. YBX1 has been shown to interact with RBBP6, PCNA, ANKRD2, SFRS9, CTCF and P53. These genes and proteins are therefore considered as RNA splicing for the interpretation of the protein application.

Preferred prognostic determinants of the present invention include ACTN1, FUS, SMAD2, HOXB13, DERL1, pS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, AKAP8, DIABCD, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. More preferred prognostic determinants of the present invention include ACTN1, FUS, SMAD2, DERL1, pS6, YBX1, SMAD4, VDAC1, DCC, CUL2, PLAG1 and PDSS2. These 12 more preferred biomarkers are shown in more detail in Table 1 below.

  As used herein, the term “ACTN1” refers to actinin, α1. ACTN1 may also be what is known as actinin α1, α-actinin cytoskeletal isoform, non-muscle α-actinin-1, F-actin cross-linked protein, actinin 1 smooth muscle or α-actinin-1. This is an F-actin cross-linked protein that can anchor actin into various intracellular structures. For example, the ACTN1 protein sequence can include SEQ ID NO: 1, and the ACTN1 mRNA sequence can include SEQ ID NO: 2.

  As used herein, the term “CUL2” refers to clin-2. This is the core component of the multiple culin-RING type E3 ubiquitin-protein ligase complex. For example, the CUL2 protein sequence can include SEQ ID NO: 3 and the CUL2 mRNA sequence can include SEQ ID NO: 4.

  As used herein, the term “DCC” refers to a deleted in collective cancer. In addition, DCC is IGDCC, colorectal tumor suppressor, colorectal cancer suppressor, deleted in color cancer protein, immunoglobulin superfamily DCC subclass member 1, immunoglobulin superfamily, DCC subclass, member 1, tumor suppressor protein DCC , Netrin receptor DCC2 CRC18, and what is known as CRCR1. This is a dependent receptor. This promotes axonal outgrowth in the presence of netrin, and induces apoptosis in the absence of netrin. For example, the DCC protein sequence can include SEQ ID NO: 5 and the DCC mRNA sequence can include SEQ ID NO: 6.

  As used herein, the term “DERL1” refers to Delrin 1. DERL1 is also known as DER1, DER-1, DER1-like domain family, members, endoplasmic reticulum-related degradation protein 1, DERtri-1, FLJ13784, MGC3067, PRO2577, and Der1-like protein. Can be a thing. This is involved in ER-related degradation and reverse transport of misfolded or unfolded proteins from the ER lumen to the cytosol for proteasome degradation. For example, the DERL1 protein sequence can include SEQ ID NO: 7, and the DERL1 mRNA sequence can include SEQ ID NO: 8.

  As used herein, the term “FUS” refers to fused in sarcoma. In addition, FUS is TLS, ALS6, FUS1, oncogene FUS, oncogene TLS, translocated in liposarcoma protein, 75 kDa DNA pairing protein, amyotropic lateral sclerosis 6, hnRNP-P2, ETM4, HNRMP75, PNP, HNRMPF It may be what is known as a fusion gene in heterogeneous liposarcoma, heteronuclear ribonucleoprotein P2, RNA binding protein FUS, and POMp75. It is a member of the TET protein family involved in intracellular processes including regulation of gene expression, maintenance of genomic integrity and mRNA / microRNA processing. For example, the FUS protein sequence can include SEQ ID NO: 8 and the FUS mRNA sequence can include SEQ ID NO: 10.

  As used herein, the term “PDSS2” refers to prenyl (decaprenyl) diphosphate synthase, subunit 2. PDSS2 is also DLP1; hDLP1; COQ10D3; C6orf210; bA59I9.3; decaprenyl pyrophosphate synthetase subunit 2; decaprenyl-diphosphate synthase subunit 2; all-trans-decaprenyl-diphosphate synthase subunit 2; Decaprenyl diphosphate synthase subunit 2; decaprenyl pyrophosphate synthase subunit 2; EC 2.5.91; and chromosome 6 open reading frame 210. This is an enzyme that synthesizes the prenyl side chain of coenzyme Q or ubiquinone, which is an important element in the respiratory chain. For example, the PDSS2 protein sequence can comprise SEQ ID NO: 11 and the PDSS2 mRNA sequence can comprise SEQ ID NO: 12.

  As used herein, the term “PLAG1” refers to pleomorphic adenoma gene 1. PLAG1 can also be what is known as PSA; SGPA; ZNF912; COL1A2 / PLAG1 fusion; zinc finger protein PLAG1; and polymorphic adenoma gene 1 protein. This is a zinc finger protein with two putative nuclear localization signals. For example, the PLAG1 protein sequence can include SEQ ID NO: 13, and the PLAG1 mRNA sequence can include SEQ ID NO: 14.

  As used herein, the term “RpS6” refers to ribosomal protein S6. RpS6 may also be known as S6; phosphoprotein NP33; and 40S ribosomal protein S6. This is a cytoplasmic ribosomal protein that is a component of the 40S ribosomal subunit. For example, the RpS6 protein sequence can comprise SEQ ID NO: 15 and the RpS6 mRNA sequence can comprise SEQ ID NO: 16.

  As used herein, the term “SMAD2” refers to SMAD family member 2. SMAD2 is also known as JV18; MADH2; MADR2; JV18-1; hMAD-2; hSMAD2; SMAD family member 2; SMAD, mothers against DPP homolog 2 (Drosophila); -And Mad related protein 2; MAD homolog 2; Mad related protein 2; mothers against DPP homolog 2; It is a member of the Smad family of proteins that are signaling and transcriptional regulators that mediate many signaling pathways such as the TGF-β pathway, cell proliferation processes, apoptotic processes and differentiation processes. For example, the SMAD2 protein sequence can include SEQ ID NO: 17, and the SMAD2 mRNA sequence can include SEQ ID NO: 18.

  As used herein, the term “SMAD4” refers to SMAD family member 4. In addition, SMAD4 is, JIP; DPC4; MADH4; MYHRS; deleted in pancreatic carcinoma locus 4; mothers against decapentaplegic homolog 4; mothers against decapentaplegic, Drosophila, homolog, 4; deletion target in pancreatic carcinoma 4; SMAD, mothers against DPP homolog 4 MAD homolog 4; hSMAD4; MAD, mothers Against detentative homolog 4 (Drosophila); mothers again DPP homolog 4; and SMAD, mothers agai It may be those that are known as st DPP homolog 4 (Drosophila). It is a member of the Smad family protein and can form homomeric and heteromeric complexes with other active Smad proteins, which then accumulate in the nucleus and regulate transcription of the target gene. For example, the SMAD4 protein sequence can include SEQ ID NO: 19, and the SMAD4 mRNA sequence can include SEQ ID NO: 20.

  As used herein, the term “VDAC1” refers to voltage-dependent anion channel 1. VDAC is also known as VDAC-1; PORIN; MGC111064; mitochondrial outer membrane protein porin 1; voltage-dependent anion-selective channel protein 1; protoplasmic porin; VDAC; porin 31HL; hVDAC1; and porin 31HM It can be. This is a voltage-dependent anion channel protein that is a major component of the mitochondrial outer membrane. This can facilitate the exchange of metabolites and ions across the mitochondrial outer membrane and can regulate mitochondrial function. For example, the VDAC1 protein sequence can comprise SEQ ID NO: 21 and the VDAC1 mRNA sequence can comprise SEQ ID NO: 22.

  As used herein, the term “YBX1” refers to Y box binding protein 1. YBX1 is YB1; BP-8; YB-1; CSDA2; NSEP1; MDR-NF1; NSEP-1; Nuclease sensitive element binding protein 1; DBPB; Enhancer Factor I subunit A; CBF-A3; CCAAT binding transcription factor I subunit A; DNA binding protein B; Y box transcription factor; CSDB; Y box binding protein 1; major histocompatibility complex, class II, Y box binding protein I; and nuclease sensitive element binding protein Can be known as 1. This mediates the regulation of pre-mRNA alternative splicing. For example, it can bind to a splice site within the pre-mRNA and regulate splice site selection. It can also bind to and stabilize cytoplasmic mRNA. For example, the YBX1 protein sequence can include SEQ ID NO: 23, and the YBX1 mRNA sequence can include SEQ ID NO: 24.

  Another biomarker referred to herein is HSPA9. As used herein, the term “HSPA9” refers to 70 kDa heat shock protein 9 (Mortarin). HSPA9 may also be what is known as CSA; MOT; MOT2; GRP75; PBP74; GRP-75; HSPA9B; MTHSP75; The Entrez Gene ID of human HSPA9 is 3313. The human HSPA9 mRNA sequence is shown in NM_004134.6 (SEQ ID NO: 26). The human HSPA9 protein sequence is shown in NP_004125.3 (SEQ ID NO: 25). For example, the HSPA9 protein sequence can comprise SEQ ID NO: 25. For example, the HSPA9 mRNA sequence can comprise SEQ ID NO: 26.

  The sequences shown herein are merely exemplary. The biomarkers of the present invention can be any form and variant of any of the specifically described biomarkers, including, but not limited to, polymorphic variants or alleles composed of any biomarker as a constituent subunit of a complete assembly structure Includes variants, isoforms, variants, derivatives, precursors (eg, nucleic acids and proproteins), cleavage products and structures.

Construction of a biomarker panel As mentioned above, the possibility that PD may correlate with cancer prognosis can be amplified by using these in combination. Thus, the biomarker panel of the present invention may be constructed with two or more of the PDs described herein. An example of a biomarker panel of the present invention comprises 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 biomarkers, each of which is independently at least 1 At least one ubiquitination gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; at least one terpenoid skeleton biosynthesis gene or protein; At least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-gated anion channel gene or protein; and at least one RNA splicing gene or tamper It can be one that is selected from the quality. Preferably, the biomarker panel comprises 6, 7, 8 or 9 biomarkers, most preferably 7 biomarkers.

  An example of a preferred biomarker panel of the present invention comprises 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 biomarkers, each of which is independently ACTN1 , FUS, SMAD2, HOXB13, DERL1, pS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, AKAP8, DIABLO, CD75, LATS2, EDEC1, HMO7, DEC1, HMO7 , MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. An example of a preferred biomarker panel of the present invention comprises 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 biomarkers, each of which is independently ACTN1 , FUS, SMAD2, DERL1, pS6, YBX1, SMAD4, VDAC1, DCC, CUL2, PLAG1 and PDSS2. Preferably, the biomarker panel comprises 6, 7, 8 or 9 biomarkers, most preferably 7 biomarkers. The exact combination and weighting of biomarkers can vary depending on the prognostic information to be obtained.

The following biomarker combinations are envisioned:
1. PD1 and PD2, where PD1 and PD2 are different;
2. PD1, PD2 and PD3, where PD1, PD2 and PD3 are different;
3. PD1, PD2, PD3 and PD4, where PD1, PD2, PD3 and PD4 are different;
4). PD1, PD2, PD3, PD4 and PD5, where PD1, PD2, PD3, PD4 and PD5 are different;
5. PD1, PD2, PD3, PD4, PD5 and PD6, where PD1, PD2, PD3, PD4, PD5 and PD6 are different;
6). PD1, PD2, PD3, PD4, PD5, PD6 and PD7, where PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are different;
7). PD1, PD2, PD3, PD4, PD5, PD6, PD7 and PD8, where PD1, PD2, PD3, PD4, PD5, PD6, PD7 and PD8 are different;
8). PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8 and PD9, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8 and PD9 are different;
9. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9 and PD10, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9 and PD10 are different;
10. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10 and PD11, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10 and PD11 are different is there;
11. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12 Are different;
In this case, PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12 are each independently ACTN1, FUS, SMAD2, HOXB13, DERL1, pS6, FAK1, YBX1, SMAD4 , VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, and PR3 The

The following biomarker combinations are preferred:
1. PD1 and PD2, where PD1 and PD2 are different;
2. PD1, PD2 and PD3, where PD1, PD2 and PD3 are different;
3. PD1, PD2, PD3 and PD4, where PD1, PD2, PD3 and PD4 are different;
4). PD1, PD2, PD3, PD4 and PD5, where PD1, PD2, PD3, PD4 and PD5 are different;
5. PD1, PD2, PD3, PD4, PD5 and PD6, where PD1, PD2, PD3, PD4, PD5 and PD6 are different;
6). PD1, PD2, PD3, PD4, PD5, PD6 and PD7, where PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are different;
7). PD1, PD2, PD3, PD4, PD5, PD6, PD7 and PD8, where PD1, PD2, PD3, PD4, PD5, PD6, PD7 and PD8 are different;
8). PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8 and PD9, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8 and PD9 are different;
9. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9 and PD10, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9 and PD10 are different;
10. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10 and PD11, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10 and PD11 are different is there;
11. PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12, where PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12 Are different;
In this case, PD1, PD2, PD3, PD4, PD5, PD6, PD7, PD8, PD9, PD10, PD11 and PD12 are each independently ACTN1, FUS, SMAD2, DERL1, pS6, YBX1, SMAD4, VDAC1, DCC. , CUL2, PLAG1 and PDSS2.

  If desired, the biomarker combination includes at least ACTN1, YBX1, SMAD2, and FUS. Alternatively, the biomarker combination comprises (1) at least ACTN1, YBX1 and SMAD2; (2) at least ACTN1, YBX1 and FUS; (3) at least ACTN1, SMAD2 and FUS; or (4) at least YBX1, SMAD2 and FUS. Is included. An example of a preferred combination of biomarkers is shown in Table 6 disclosed in US Provisional Patent Application No. 61 / 792,003 (filed Mar. 15, 2013, the entire contents of which are incorporated herein by reference). It is a thing.

Tissue Sample The tissue sample used in the methods of the present invention can be a tumor sample obtained by biopsy (eg, a prostate tumor sample). An insurance provider may order a biopsy (eg, a prostate biopsy) if prostate cancer is suggested by results of an initial test, such as a prostate specific antigen (PSA) blood test or a digital rectal exam (DRE) . In order to obtain a prostate biopsy, insurance organizations can use a fine needle to take several tissue samples (also called “core” samples) from the glandular portion of the prostate (see also discussion below) . The tissue sample for the method of the present invention may also be obtained by surgery (eg, prostatectomy) performed by a urologist or a surgical support robot. The tissue sample obtained by surgery may be the entire prostate or part of the prostate and may contain one or more lymph nodes. In one embodiment, the tissue sample can be block formalin fixed paraffin embedded (FFPE). A section can then be cut from the FFPE block and placed on the slide by any suitable means. Slides containing samples from many tumors or patients may be combined into one batch as a tissue microarray (TMA) for processing. Frozen tissue can be used as well. Add to batch the appropriate control slide or control core, eg prepared from cell lines with extensive expression of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1 Also good.

  A set of control cell lines showing high, intermediate and low levels of expression for each biomarker (eg, ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1) Can be selected. These cell lines can then be fixed in formalin, processed and incorporated into paraffin blocks using standard histological techniques. A cell line control TMA can be established by placing the core of each cell line paraffin block into a new acceptor block. The cell line control TMA can be sectioned and the resulting sections can be stained in parallel with the patient tissue sample. Such a cell line control TMA can be used as a reference point for an immunostaining quantitative assay that measures biomarker expression in a patient tissue sample, since the cell line provides a source of uniform and reproducible biomarker expression. By comparing with quantitative control levels over time, the user can see if the instrument tends to be out of the calibration standard. Also, if necessary, the user may standardize patient samples against control values for absolute quantification between batches.

Measurement of Biomarkers The biomarkers of the present invention can be measured in various forms. For example, the biomarker level is at the genomic DNA level (eg, measuring copy number, heterozygosity, deletion, insertion or point mutation), at the mRNA level (eg, measuring transcript level or transcript location), It can be measured at the protein level (eg, protein expression level, post-translational modification quantification or activity level) or at the metabolite / analyte level. Methods for measuring biomarker level genomic DNA, mRNA, protein and metabolite / analyte levels are known in the art. Preferably, the biomarker level is measured at the protein level in whole cells and / or cell component compartments (eg, nucleus, cytoplasm and cell membrane). Exemplary methods for measuring levels at the protein level include, but are not limited to, immunoassays such as immunohistochemical assay (IHC), immunofluorescence assay (IF), enzyme-linked immunosorbent assay (ELISA), immunoradiation Quantitative assays and immunoenzyme assays are included. In an immunoassay, for example, an antibody that binds to a biomarker or fragment thereof can be used. The antibody can be monoclonal, polyclonal, chimeric or humanized. The antibody may be bispecific. In addition, antigen-binding fragments of complete antibodies, such as single chain antibodies, Fv fragments, Fab fragments, Fab ′ fragments, F (ab ′) ₂ fragments, Fd fragments, single chain Fv molecules (scFv), bispecific single Chain Fv dimers, nanobodies, diabodies, domain deleted antibodies, single domain antibodies and / or oligoclonal mixtures of two or more specific monoclonal antibodies can also be used.

  For example, a tissue sample, such as the biopsy slide described above, can be assayed, eg, in an immunohistochemical (IHC) assay, to determine the level of an appropriate biomarker. In IHC assays, antibodies detectably labeled against various biomarkers can be used for staining of prostate tissue samples and the binding level can be displayed, for example, by fluorescence or luminescence. Colorimetric dyes (eg DAB, Fast Red) can be used as well. In one embodiment, the prostate tissue slides use one or more of the antibodies that bind to ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1, respectively. Stained. The antibody used in the method of the present invention may be monoclonal or polyclonal. An antigen-binding portion of a complete antibody or any other molecular entity (eg, peptidomimetics and aptamers) that can specifically bind to a biomarker may also be used.

Other methods for measuring biomarkers at the protein level include, for example, chromatography, mass spectrometry, Luminex xMAP Technology, microfluidic chip-based assays, surface plasmon resonance, sequencing, Western blot analysis, aptamer binding, molecular Printing methods, peptidomimetics, peptide bonds based on affinity, chemical bonds based on affinity, or combinations thereof. Also, AQUA® (see, eg, US Pat. Nos. 7,219,016 and 7,709,222; Camp et al., Nature Medicine, for measuring total intracellular and / or intracellular component levels of biomarkers. 8 (11): 1323-27 (2002)) and Definiens TissueStudio ^™ (eg, US Pat. Nos. 7,873,223, 7,801,361, 7,467,159 and 7,146,380, and Methods such as Baatz et al., Comb Chem High Throughput Screen, 12 (9): 908-16 (2009)) can also be used.

  In some embodiments, the measured biomarker levels are normalized to a normalizing protein, such as an expression product of a housekeeping gene, such as GAPDH, Cyn1, ZNF592, or an actin expression product, to remove sources of variation. It is burned. Normalization methods are well known in the art. See, for example, Park et al., BMC Bioinformatics. 4:33 (2003).

Defining the region of interest In order to improve the accuracy of the assay, it may be desirable to define the region of interest and quantify the biomarker only within the region of interest. The region of interest can be defined by applying a “tumor mask” to the sample so that only the biomarker level in the tumor region is measured. “Tumor mask” refers to a combination of biomarkers that allows the identification of a tumor region within a target tissue. For example, prostate cancer typically expresses epithelial markers such as cytokeratin 8 (CK8 or KRT8) and cytokeratin 18 (CK18 or KRT18), but a prostate basal marker such as cytokeratin 5 (CK5 or KRT5). ) Is a non-expressing cancer. Thus, a “tumor mask” in prostate cancer may involve the use of a mixture of antibodies that specifically bind to these markers. The inventors have also surprisingly found that TRIM29, a tumor marker for some other cancers, is a basal marker rather than a tumor marker in prostate tissue; thus, anti-TRIM29 antibodies are also Can be used for prostate tumor masks. For example, a prostate tumor mask useful in the present invention may comprise a mixture of anti-CK5, anti-CK8, anti-CK18 and anti-TRIM29 antibodies, in which case the prostate tumor region is bounded by anti-CK8 and anti-CK18 antibodies. Region of prostate tissue that is not bounded by anti-CK5 and anti-TRIM29 antibodies. A prostate tumor region may be defined as a prostate tissue region bounded by either anti-CK8 or anti-CK18 antibody, preferably both. Similarly, a prostate tumor region may be defined as a prostate tissue region that is not bordered with anti-CK5 antibody or borderlined with anti-TRIM29 antibody. Preferably, the prostate tumor area is not bounded by either anti-CK5 or anti-TRIM29 antibodies. A basal prostate tumor region may be defined as a prostate tissue region bounded by either anti-CK5 or anti-TRIM2 antibody, preferably both. Preferably, the basal tumor region is not bounded by either anti-CK8 or anti-CK18 antibodies. Alternatively, other combinations of epithelial and basal markers may be used, for example, epithelial cell ESA antibody and basal cell p63 antibody. For cancers other than prostate cancer, other marker combinations that allow the identification of the tumor area may be used, eg, the S100 marker specific for malignant melanoma.

  Thus, according to one aspect of the invention, for defining a region of interest in a tissue sample comprising contacting the tissue sample with one or more first reagents for specifically identifying the region of interest. Provide a way. The region of interest may include cancer cells such as prostate cancer cells. To identify prostate cancer cells, the one or more first reagents may comprise an anti-cytokeratin 8 antibody, an anti-cytokeratin 18 antibody, or both. The method further includes one or more second reagents for specifically identifying a tissue sample region to be excluded from the region of interest, eg, a non-cancerous cell, and the tissue sample to the region to be excluded. And defining by contacting with. For example, to exclude non-cancerous prostate basal cells, the one or more second reagents may include an anti-cytokeratin 5 antibody, an anti-TRIM29 antibody, or both.

  In order to be able to measure biomarkers in cell component regions, such as the nucleus, cytoplasm and cell membrane, it is necessary to use specific markers for the regions. Cytokeratins 8 and 18 used for identification of epithelial regions provide cytoplasm-specific and membrane-specific staining patterns and can therefore be used to define this subcellular localization. To identify the nuclear region of the cell, the prostate tissue sample can be stained with a nuclear specific fluorescent dye, such as DAPI or Hoechst 33342.

  After appropriate staining, the biopsy slide can be processed such that the signal for detection is preserved, for example, by applying an anti-fading reagent and / or a coverslip on the slide. The slide is then stored in the imaging device and can be read. The images so obtained can then be processed and biomarker expression can be quantified. This process is also referred to as quantitative multiplex immunofluorescence collection (QMIF collection).

  The multiplex in situ proteomics technology of the present invention provides several advantages over conventional genetic platforms where gene expression is measured rather than protein expression / activity. First, the use of a tumor mask makes it possible to obtain marker information only from tumor tissue that is not “diluted” by normal tissue, thus improving examination accuracy. The technique also allows for marker quantification in various regions of tumor tissue that are known to be fairly heterogeneous. Read-out information from the most aggressive areas of the tumor provides a more accurate view of the patient's clinical outcome and is thus more useful in assisting the physician in determining the best course of treatment for the patient. In addition, the multivariate diagnostic method of the present invention is designed to predict outcome even in a smaller representative tumor region, alleviating problems caused by random sampling errors. Furthermore, the use of the activated antibody and the localization of the marker in the cellular component enables the quantitative determination of the functionally active marker, further improving the test accuracy.

Data Processing Images obtained from immunofluorescence of tumor samples can be generated using algorithms suitable for automated quantitative analysis of data collected from images (eg, pattern recognition software or other using algorithms developed using Definiens Developer XD ^™). In some embodiments of the present invention, such an algorithm can be exported to ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 in some embodiments of the present invention. And the presence and / or level of antibody staining for one or more of YBX1 and the presence of CK8 and CK18 staining and CK5 and TR. An algorithm for focusing on the tumor area defined by the absence of IM29 staining may be used, hi some embodiments, ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4 , An algorithm for creating a heat map of the most aggressive area for one or more of VDAC1 and YBX1 may be used, and an algorithm for measuring tumor volume may be used.

  Data obtained by image processing of the tissue sample is used to calculate the risk score. The risk score can measure the aggressiveness of a tumor (eg, a prostate tumor). For example, the risk score can predict the probability that a tumor (eg, prostate tumor) is actively progressing at the time of diagnosis or is indolent / dormant. The risk score can also predict the probability that a tumor (eg, prostate tumor) is progressing at a later time after the diagnosis. The risk score also determines the fatal outcome / disease-specific mortality (DSD) of the cancer (eg, prostate cancer), ie the probability that a patient with the tumor will die of the cancer (eg, predicted years of survival), or the tumor (eg, The risk of progression or metastasis of prostate tumors may also be predicted. These probabilities can be obtained by evaluating the model at marker values measured in the sample / classifier trained to predict this risk. Several probabilistic binomial classifiers can be used and are known to those skilled in the art (such as random forest or logistic regression). In the examples shown below, logistic regression was used. The risk score can also be used to detect cells that have the potential for metastasis in a tumor tissue sample. Also, other diagnostic results or cancer parameters in the risk score, eg digital rectal examination (DRE) results, prostate specific antigen (PSA) levels, PSA kinetics, Gleason score, tumor progression classification, tumor size, onset age and lymph A clause state may be incorporated. The risk score can be communicated to the insurance medical institution and / or patient and used to determine a treatment regimen for the patient (eg, surgery).

Clinical Applications The diagnostic methods of the present invention are useful for insurance medical institutions to determine the most appropriate treatment for cancer patients (eg, prostate cancer patients). If an insurance medical institution suspects a patient's cancer (eg, prostate cancer) based on medical history, DRE and / or PSA levels, the institution may order a biopsy (eg, prostate biopsy). To perform a biopsy, a general practitioner or urologist will use a transrectal ultrasound (TRUS) guide core needle to produce multiple (eg, 8 to 18) core samples (each about ½ inch each). Long and 1/16 inch wide) can be taken. If cancerous cells are found by morphological examination, further examinations (eg, imaging examinations such as bone scans, CT scans and MRI Prostatint ^™ Scan) can be performed to assist in the classification of cancer progression. The diagnostic methods of the invention can then be performed to further predict cancer aggressiveness, risk of progression or outcome. If the method predicts 1) active progression of the tumor; 2) high risk of progression; or 3) fatal outcome, insurance medical institutions may decide to use aggressive treatment. For example, in addition to prostatectomy, doctors may provide radiation therapy (eg, external radiation, proton therapy and brachytherapy), hormone therapy (eg, testicular incision, LHRH agonist or antagonist formulation and antiandrogen), chemotherapy, As well as other suitable treatments such as Siplecel-T (PROVENGE®) therapy, cryosurgery and high intensity laser therapy. However, if the patient is prognostically diagnosed with indolent PCA, it is referred to as active monitoring and can be subjected to repeated biopsies and does not require radical treatment.

  Accordingly, one aspect of the present invention provides a method for predicting the prognosis of a cancer patient. The method comprises measuring, in a sample taken from a patient, at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; At least one terpenoid skeleton biosynthesis gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion A channel gene or protein; and can comprise two or more PD levels selected from at least one RNA splicing gene or protein; in this case, the level measurement is Show the prognosis of the patient. If necessary, the two or more PDs are ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, It is selected from the group consisting of AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Preferably, the two or more PDs are selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1. The method may further include collecting a sample from the patient. The prognosis may be that the cancer is an aggressive form of cancer, the patient is at risk of having an aggressive form of cancer, or the patient is at risk of having a cancer-related lethal outcome. The cancer can be prostate cancer.

  According to another aspect of the invention, obtaining a tissue sample from a patient; and in said sample, at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor At least one DNA repair gene or protein; at least one terpenoid skeleton biosynthesis gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; Measuring two or more PD levels selected from one voltage-gated anion channel gene or protein; and at least one RNA splicing gene or protein. ; In this case, the level measurement indicates that the patient is in need of adjuvant therapy provides methods for identifying a cancer patient in need of adjuvant therapy. If necessary, the two or more PDs are ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, It is selected from the group consisting of AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Preferably, the two or more PDs are selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1.

  According to a further aspect of the invention, at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; At least one terpenoid backbone biosynthetic gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion channel gene or protein; and at least one Measuring the level of two or more PDs selected from the group consisting of two types of RNA splicing genes or proteins; and level progression to actively progress the patient If it has a risk of having a risk or cancer-related fatal outcome of cancer or cancer progression it has been shown, comprising treating the patient with adjuvant therapy, provides a method for treating cancer patients. If necessary, the two or more PDs are ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, It is selected from the group consisting of AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Preferably, the two or more PDs are selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1. Alternatively, the method identifies a patient having at least two types of PD level changes, wherein the level changes are of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2 and VDAC1. One or more up-regulations and one or more down-regulations of ACTN1, RpS6, SMAD4 and YBX1); and treating the patient with adjuvant therapy Including. The patient can be a patient with prostate cancer.

  Adjuvant therapy can be selected from the group consisting of radiation therapy, chemotherapy, immunotherapy, hormone therapy and targeted therapy. In some embodiments, the targeted therapy targets one component of a signal transduction pathway in which one or more of the selected PDs are one component, in which case the targeted component Is different from the selected PD. Alternatively, the targeted therapy is one that targets one or more of the selected PDs. The patient may be a patient who has been subjected to standard care treatments such as surgery, radiation, chemotherapy or androgen removal.

  In a further aspect of the invention, preparing a cell expressing PD selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1; Contacting the candidate compound and determining whether the candidate compound modifies the expression or activity of the selected PD; wherein the observed observation in the presence of the compound is that the compound is cancerous Provided are methods for identifying compounds that can reduce the risk of progression, or that can reduce or delay the progression of cancer, that can reduce the risk of cancer progression, or that can delay or delay the progression of cancer.

  According to another aspect of the invention, at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one DNA repair gene or protein; At least one terpenoid backbone biosynthetic gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion channel gene or protein; and at least one Measuring the level of two or more PDs selected from the group consisting of different RNA splicing genes or proteins; and modulating the level of the selected PD Comprising administering an agent, it provides a method for treating cancer patients. If necessary, the two or more PDs are ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, It is selected from the group consisting of AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Preferably, the two or more PDs are selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1. Alternatively, the method identifies a patient having at least two types of PD level changes, wherein the level changes are of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2 and VDAC1. Selected from the group consisting of one or more up-regulation of and one or more down-regulation of ACTN1, RpS6, SMAD4 and YBX1); and at least one level of said PD Administering a modulating agent.

  In any of the above methods, at least 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 levels of PD can be measured. As required, the levels of six types of PD consisting of PD1, PD2, PD3, PD4, PD5 and PD6 are measured, where PD1, PD2, PD3, PD4, PD5 and PD6 are different and independent. ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, AKAP8, DIABLO, CD75, LATSL, E75 Selected from the group consisting of CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5 and pPRAS40. Preferably, PD1, PD2, PD3, PD4, PD5 and PD6 are different and independently comprise the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1 More selected.

  If necessary, the levels of seven types of PD consisting of PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are measured, where PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are different. Yes, independently, ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, AKAP8, DIABLO, CD75, TS It is selected from the group consisting of LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5 and pPRAS40. Preferably, PD1, PD2, PD3, PD4, PD5, PD6 and PD7 are different and independently from ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1. Selected from the group consisting of The method is further selected from the group consisting of HOXB13, FAK1, COX6C, FKBP5, PXN, AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Measuring the level of one or more PDs may be included.

  The level measurement of the at least one PD can be up-regulated relative to the reference value. Preferably, the up-regulated PD is selected from the group consisting of CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, SMAD2 and VDAC1. Furthermore, the level measurement of at least one PD may be down-regulated compared to the reference value. Preferably, the down-regulated PD is selected from the group consisting of ACTN1, RpS6, SMAD4 and YBX1. Accordingly, the level measurement of at least one PD may be up-regulated relative to the reference value, and the measurement of at least one PD may be down-adjusted relative to the reference value. The reference value can be a measure of the level of PD in non-cancerous cells.

  Any of the above methods involves measuring the genomic DNA level, mRNA level or protein level of each PD. For example, the method can include contacting the sample with an oligonucleotide, aptamer or antibody specific for each PD. The level of PD can be measured separately or in parallel, for example using a multiplex reaction. Preferably, the protein level of each PD is measured. For example, antibodies or antibody fragments can be used to measure protein levels by immunohistochemistry or immunofluorescence. If more than one PD is measured in a single sample, each antibody or fragment thereof can be labeled with a different fluorophore or conjugated with a different fluorophore. Signals from the different fluorophores can be detected in parallel by an automated imaging device.

  PD protein levels may be measured in specific cell component compartments. For example, DAPI staining can be used to identify the nucleus of each cell and thus the amount of each PD in the nucleus and / or cytoplasm can be measured.

  Similarly, PD levels may be measured only in defined regions of interest. In cancer, for example, cancer cells can be included in the region of interest while non-cancerous cells can be excluded from the region of interest. In the prostate, cancer cells express cytokeratin-8 and cytokeratin-18, and basal (non-cancer) cells express cytokeratin-5 and TRIM29. Thus, the region of interest can be defined by staining for anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody, and can be further defined by the lack of staining for anti-cytokeratin 5 antibody and anti-TRIM29 antibody. The exclusion region can be defined by staining for anti-cytokeratin 5 antibody and anti-TRIM29 antibody, and can be further defined by the lack of staining for anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody.

  In any of the above methods, the sample is a solid tissue sample or a blood sample, preferably a solid tissue sample. Solid tissue samples are removed by formalin fixed paraffin embedded tissue samples, snap frozen tissue samples, ethanol fixed tissue samples, tissue samples fixed with organic solvents, tissue samples fixed with plastic or epoxy, cross-linked fixed tissue samples, surgery Tumor sample or biopsy material sample, such as core biopsy material, and excised tissue biopsy material or incision tissue biopsy material. Preferably, the sample is a cancerous tissue sample. The sample can be a prostate tissue sample, eg, a formalin fixed paraffin embedded (FFPE) prostate tumor sample. Accordingly, the above method further comprises contacting a cross-section of the FFPE prostate tumor sample with an anti-cytokeratin 8 antibody, an anti-cytokeratin 18 antibody, an anti-cytokeratin 5 antibody and an anti-TRIM29 antibody, wherein the measuring step May be performed in a region within the cross section that is demarcated by anti-cytokeratin 8 antibody and anti-cytokeratin 18 antibody, but not delimited by anti-cytokeratin 5 antibody and anti-TRIM29 antibody.

  Any of the above methods may further comprise measuring at least one standard parameter associated with cancer. Standard parameters include, but are not limited to, Gleason score, tumor progression classification, tumor grade, tumor size, tumor appearance characteristics, tumor location, tumor growth, lymph node status, tumor thickness (brace Low score), ulceration, age of onset, PSA level, and PSA kinetics.

Additional Prognostic Factors The biomarker panels of the present invention can be used with additional biomarkers, clinical parameters or conventional laboratory risk factors that are known to exist or are associated with the subject's clinical outcome. One or more clinical parameters may be used in the practice of the invention as biomarkers to be included in a formulation or as a pre-selection criteria that defines the appropriate population to be measured using a specific biomarker panel and formulation . One or more clinical parameters are also useful for biomarker normalization and pretreatment, or biomarker selection, panel construction, formulation type selection, and post-treatment formulation results obtain. A similar approach can be performed using conventional test value risk factors. Clinical parameters or conventional laboratory risk factors are clinical signs that are typically evaluated in a clinical laboratory and used in conventional comprehensive risk assessment algorithms. Clinical parameters of tumor metastasis or conventional laboratory risk factors include, for example, tumor progression classification, tumor malignancy, tumor size, tumor appearance characteristics, tumor location, tumor growth, lymph node status, histology, Tumor thickness (Breathrow score), ulceration, growth index, tumor infiltrating lymphocytes, age of onset, PSA level, or Gleason score may be mentioned. Other conventional laboratory value risk factors for tumor metastasis are known to those skilled in the art.

  In some embodiments, the biomarker score obtained by the method of the present invention can be used in conjunction with the Gleason score to obtain better prediction results. The Gleason score is given to prostate cancer based on the microscopic appearance of prostate tissue and is used clinically to predict the prognosis of PCA. To obtain a Gleason score, prostate tissue samples can be stained with hematoxylin and eosin (H & E) and examined by a pathologist under a microscope. The prostate tumor pattern of the sample is classified on a 1-5 scale, with 5 being the least differentiated and invasive. The tumor Gleason score is constructed by adding the classification of the second most common pattern (less than 50% but more than 5%) to the classification of the most common pattern (more than 50% of the tumor). Scores 2-6 indicate low grade PCA and the risk of recurrence is low. Score 7 (3 + 4 or 4 + 3) indicates moderate grade PCA, with a moderate risk of recurrence, where score 4 + 3 is worse than score 3 + 4. Scores 8-10 indicate high-grade PCA and a high risk of recurrence. The risk score determined by the methods described herein can be used in conjunction with the Gleason score to improve the likelihood of predicting the Gleason score. For example, an intermediate Gleason score of 7 (3 + 4) does not provide a good prediction of a patient's PCA recurrence risk. However, taking into account the risk score calculated by the method described herein improves the predictive power of the intermediate Gleason score.

Kits for Detecting Biomarkers Another aspect of the invention comprises at least one cytoskeletal gene or protein; at least one ubiquitinated gene or protein; at least one dependent receptor gene or protein; at least one At least one terpenoid skeletal biosynthesis gene or protein; at least one PI3K pathway gene or protein; at least one TFG-β pathway gene or protein; at least one voltage-dependent anion A kit for measuring the level of two or more PDs selected from the group consisting of a channel gene or protein; and at least one RNA splicing gene or protein comprising: It is to a kit comprising reagents for specifically measuring the level of-option has been PD can be produced. If necessary, the two or more PDs are ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6, FAK1, YBX1, SMAD4, VDAC1, COX6C, FKBP5, DCC, CUL2, PLAG1, PDSS2, PXN, It is selected from the group consisting of AKAP8, DIABLO, CD75, LATS2, DEC1, LMO7, EIF3H, CDKN1B, MTDH2, MAOA, CCND1, HSD17B4, MAP3K5, and pPRAS40. Preferably, the two or more PDs are selected from the group consisting of ACTN1, CUL2, DCC, DERL1, FUS, PDSS2, PLAG1, RpS6, SMAD2, SMAD4, VDAC1 and YBX1.

  The reagent may measure the genomic DNA level, mRNA transcript level or protein level of the selected PD. Preferably, the reagent comprises one or more antibodies or fragments thereof, oligonucleotides or aptamers.

Biomarker Selection Method Another aspect of the present invention is a method for identifying a prognostic determinant of a target disease, comprising a biological step; a technical step; a performance step; and a verification step.

  Biological processes create candidate lists compiled for the disease of interest from publicly available data (eg, scientific literature, databases, and conference presentations); and biological relevance, in silico analysis Prioritizing the candidate list based on known expression information and the commercial availability of the required monoclonal antibodies.

  The technical process is to obtain antibodies against the candidate prognostic determinants; testing the antibodies with 3,3′-diaminobenzidine (DAB) staining in an immunohistochemical assay to assess the specificity and intensity of the staining And antibodies with sufficient specificity and intensity of staining with DAB are tested in an immunofluorescence (IF) assay to determine IF specificity, signal intensity and kinetics and identify antibodies that pass technical requirements Can include.

  A performance step contacts a mini-tissue microarray (TMA) with the antibody that passes technical requirements, where the mini-TMA includes several samples of different progression categories of the target disease; Quantifying the immunofluorescence intensity of the antibody; correlating the immunofluorescence intensity of each antibody with the prognosis of each sample in the mini-TMA; and which antibodies are univariate in the mini-TMA for correlation with the prognosis of the target disease This may include examining what is showing performance. If necessary, the performance process may further contact a larger TMA with the antibody that passes the technical requirements (where the larger TMA is associated with some of the different progression categories of the subject disease). Quantifying the immunofluorescence intensity of each antibody; correlating the immunofluorescence intensity of each antibody with the prognosis of each sample within the larger TMA; and correlating with the prognosis of the target disease To find out which antibodies show univariate performance in the larger TMA. In some embodiments, the performance step further includes performing a bioinformatics analysis to identify antibody combinations against PD that correlate with the prognosis of the target disease.

  The verification step comprises collecting a tissue sample from a patient suffering from the target disease; contacting the tissue sample with an antibody against PD or a combination of antibodies against the target disease; immunofluorescence intensity of each antibody or combination of antibodies And correlating the immunofluorescence intensity of each antibody or combination of antibodies with the prognosis of the subject in the subject disease.

Computer System Examples Various aspects and functions described herein in accordance with this disclosure may be implemented in one or more computer systems as hardware, software, or a combination of hardware and software. There are many examples of computer systems currently in use. Examples include network appliances, personal computers, workstations, mainframes, networked clients, servers, media servers, application servers, database servers, web servers, and virtual servers, among others. Other examples of computer systems may include mobile computing devices such as mobile phones and personal digital assistants, and network equipment such as load balancers, routers and switches. Further, aspects in accordance with the present disclosure may be located in a single computer system or distributed among multiple computer systems connected to one or more communication networks.

  For example, the various aspects and functions may be one or more configured such that one or more client computers are serviced, or the entire task is performed as part of a distributed system. Can be distributed among other computer systems. Further, aspects may be performed in a client-server or multi-tier system that includes components distributed among one or more server systems performing various functions. Thus, the present disclosure is not limited to execution on any particular system or group of systems. In addition, aspects may be implemented in software, hardware or firmware, or any combination thereof. Thus, although aspects in accordance with the present disclosure may be implemented using various hardware and software configurations within methods, acts, systems, placement systems and components, the present disclosure is not limited to any particular distributed structure, network, Or it is not limited to a communication protocol. Further, aspects in accordance with the present disclosure may be implemented as special program hardware and / or software.

  FIG. 26 illustrates a block diagram of a distributed computer system 100 in which various aspects and functions according to the present disclosure may be implemented. The distributed computer system 100 may include another computer system. For example, as shown, distributed computer system 100 includes three computer systems 102, 104, and 106. As shown, computer systems 102, 104, and 106 are interconnected by a communication network 108 through which data can be exchanged. Network 108 may include any communication network that can exchange data with a computer system. In order to exchange data over the network 108, the computer systems 102, 104 and 106 and the network 108 may include various methods, protocols and standards such as, among others, Token Ring, Ethernet®, Wireless Ethernet®, Bluetooth ( (Registered trademark), TCP / IP, UDP, HTTP, FTP, SNMP, SMS, MMS, SS7, JSON, XML, REST, SOAP, CORBA IIOP, RMI, DCOM, and Web Services can be used. To ensure that the data transfer is secure, the computer systems 102, 104, and 106 send data over the network 108 using various security measures, such as using TSL, SSL, or VPN, among other security techniques. Can be done. Although three computerized computer systems are illustrated in the distributed computer system 100, the distributed computer system 100 may include any number of computer systems networked using any medium and communication protocol. .

  Various aspects and functions in accordance with the present disclosure may be implemented as specialized hardware or software running on one or more computer systems, such as the computer system 102 shown in FIG. As shown, computer system 102 includes a processor 110, a memory 112, a bus 114, an interface 116 and a storage system 118. The processor 110 may include one or more microprocessors or other types of controllers and may execute a series of instructions that manipulate data. The processor 110 may be a well-known commercially available processor, such as an Intel Pentium®, Intel Atom, ARM Processor, Motorola PowerPC, SGI MIPS, Sun UltraSPARC or Hewlett-Packard PA-RISC processor, or many others. Since other processors and controllers are available, any other type of processor or controller may be used. The processor 110 may be a mobile device or smartphone processor, such as an ARM Cortex processor, a Qualcomm Snapdragon processor, or an Apple processor. As shown, the processor 110 is connected to other placement systems, for example, a memory 112 by a bus 114.

  Memory 112 may be used to store programs and data during operation of computer system 102. Thus, the memory 112 can be a relatively high performance volatile random access memory, such as dynamic random access memory (DRAM) or static memory (SRAM). However, the memory 112 may include any device for storing data, such as a disk drive or other non-volatile storage device, such as flash memory or phase change memory (PCM). In various embodiments in accordance with the present disclosure, the memory 112 may be organized into specialized, in some cases unique structures, to implement the aspects and functions disclosed herein.

  The components of computer system 102 may be coupled by interconnect elements such as bus 114. Bus 114 may include one or more physical buses (eg, a bus between components integrated within the same machine), and any communication that couples between placement systems, such as It may include specialized or standard computing bus technologies such as IDE, SCSI, PCI and InfiniBand. Thus, the bus 114 allows communications (eg, data and instructions) to be exchanged between system components of the computer system 102.

  The computer system 102 also includes one or more interface devices 116, such as input devices, output devices, and input / output integration devices. Interface device 116 may be one that accepts input, one that provides output, or both. For example, the output device can render information for external presentation. The input device may receive information from an external information source. Examples of interface devices include keyboards, mouse devices, trackballs, microphones, touch screens, printing devices, display screens, speakers, network interface cards, among others. Interface device 116 enables computer system 102 to exchange information and communicate with external entities, such as users and other systems.

  Storage system 118 may include a computer readable and computer writable non-volatile storage medium that defines a program in which instructions are stored and executed by a processor. The storage system 118 may also include information recorded on or in the medium, and this information may be processed by a program. More specifically, the information may be stored in one or more data structures that are specially configured such that storage space is maintained or execution of data exchange is increased. The instructions can be stored permanently as code signals that allow the processor to perform any of the functions described herein. Media that can be used in various embodiments can include, for example, optical disks, magnetic disks, or flash memory, among others. In operation, the processor 110 or some other controller allows data to be accessed from a non-volatile recording medium to information by the processor 110 that is faster than the storage medium included in the storage system 118, for example. It can be read into the memory 112. The memory may be located in storage system 118 or memory 112. The processor 110 can manipulate the data in the memory 112, and this data can then be copied to the media associated with the storage system 118 after processing is complete. Although various components may manage data movement between the media and the memory 112, the present disclosure is not limited thereto.

  Further, the present disclosure is not limited to a particular memory system or storage system. Although the computer system 102 illustrates, by way of example, an example type of computer system in which various aspects and functions according to the present disclosure may be implemented, aspects of the present disclosure may be implemented in the computer system illustrated in FIG. It is not limited to. Various aspects and functions according to this disclosure may also be implemented on one or more computers having different architectures or components than those shown in FIG. For example, the computer system 102 includes specially programmed special purpose hardware, such as an application specific integrated circuit (ASIC) tailored to perform the specific operations disclosed herein. There may be. In another embodiment, the same function is performed on MAC OS System X, some general purpose computing devices with Motorola PowerPC processors, and some specialized computing devices running on proprietary hardware and operating systems Can be implemented using

  The computer system 102 may include an operating system that manages at least a portion of the hardware placement included within the computer system 102. A processor or controller, such as processor 110, is a Windows®-based operating system (eg, Windows® NT, Windows® 2000 / ME, Windows® XP, Windows®), among others. 7 or Windows (registered trademark) Vista) (available from Microsoft Corporation), MAC OS System X operating system (available from Apple Computer), one of many Linux (registered trademark) operating system distributions ( For example, Enterprise Linux (registered trademark) operating system (Red Hat Inc. Available)), available from Solaris Operating System (Sun Microsystems), it may be those executing the UNIX (registered trademark) operating system (various operating systems which may be available) available from sources. The operating system can be a mobile device or a smartphone operating system, such as Windows® Mobile, Android or iOS. Many other operating systems can be used, and this embodiment is not limited to any particular operating system. The computer system 102 may include a virtualization feature that is a host for an operating system within a virtual machine (eg, a pseudo physical machine). Various components of the system architecture can reside in individual case operating systems within separate “virtual machines” and thus can be executed somewhat isolated from one another.

  The processor and operating system together define a computing platform on which application programs can be written in a high level programming language. Such a component application may be executable intermediate code (eg, C # or JAVA® bytecode) and may use a communication network (eg, Internet) using a communication protocol (eg, TCP / IP). ) May be interpreted execution code communicated above. Similarly, functionality according to aspects of the present disclosure may be implemented using an object-oriented programming language, such as SmallTalk, JAVA, C ++, Ada, or C # (C-Sharp). Other object-oriented programming languages can also be used. Alternatively, procedural, script or logical programming languages may be used.

  In addition, functions according to various aspects of the present disclosure may render aspects of a graphical user interface or perform other functions when viewed in a non-programming environment (eg, HTML, XML, or browser program window). Documents created in other formats can be implemented. Moreover, various embodiments in accordance with aspects of the disclosure may be implemented as programmed or unprogrammed placements, or any combination thereof. For example, a web page can be implemented using HTML, but a data object called from within a web page can be written in C ++. Thus, the present disclosure is not limited to a particular programming language and any suitable programming language can also be used.

  Functions may be performed outside the scope of this disclosure in a computer system included in one embodiment. For example, aspects of the system can be obtained using existing products, such as the Google search engine, the Yahoo search engine (available from Yahoo! (Sunnyvale, California)) or the Bing search engine (Microsoft (Seattle, Washington)). ) Or the like. Aspects of the system are database management systems such as SQL Server (available from Microsoft (Seattle, Wash.)); Oracle Database (from Oracle (Redwood Shores, California)); and MySQL (Sun Ms. Or may be implemented in integrated software, such as WebSphere middleware (IBM (from Armonk, New York)). However, for example, a computer system running SQL Server may support both aspects according to the present disclosure and databases of various applications that are not within the scope of the present disclosure.

  Also, the methods described herein may be incorporated into other hardware and / or software products, such as web publishing products, web browsers, or internet marketing or search engine optimization tools.

  Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Exemplary methods and materials are described below, but methods and materials similar or equivalent to those described herein can also be used in the practice or testing of the present invention. All publications and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification (including definitions) will control. Although several documents are listed herein, this list does not constitute an admission that any of these documents forms part of the well-known general knowledge in the art. Throughout this specification and embodiments, the word “comprise” or ending variants such as “comprises” or “comprising” include the stated integer or group of integers and any other integer It will be understood that it is not meant to exclude an integer group. The materials, methods, and examples are illustrative only and not intended to be limiting.

Examples Further details of the invention are described in the following non-limiting examples. The examples illustrate some preferred embodiments of the present invention, but are given for illustrative purposes only and should not be construed to limit the appended embodiments or the claims. Please understand that. From this disclosure and the examples, those skilled in the art can ascertain certain features of the invention, and to implement it without departing from the spirit and scope of the invention to adapt it to various uses and conditions. Examples can be changed and modified.

Example 1 Generation of Tumor Microarrays In the experiments described in Examples 2-4, four different tumor microarrays (TMA) were used: cell line control TMA, mini TMA, high reading Gleason TMA and low reading Gleason TMA. .

A. Generation of cell line control TMA A set of cell line controls was selected to examine the reliability and reproducibility of the multiplex immunofluorescence assay. These cell lines had a range of expression levels against tumor markers that could be analyzed in a multiplex immunofluorescence assay. The cell lines and their biomarker expression levels are shown in Table 2 below.

  Selected cell lines were cultured under standard conditions and treated with PI3K kinase inhibitors (see Table 2) when necessary. Cells were washed with PBS, fixed directly on the plate with 10% formalin for 5 minutes, scraped and collected in fixative, and fixation was continued at room temperature for a total of 1 hour. Cells were then spun down and washed twice with PBS. The cell pellet was resuspended in 70 ° C. warm Histogel and quickly spun down into an Eppendorf tube to form a concentrated cell-Histogel pellet. This pellet was embedded in paraffin into a standard paraffin block and used as a donor block for tissue microarray (TMA) construction.

  The TMA block was made using a modified agarose block procedure (Yan et al., J Histochem Cytochem 55 (1): 21-24 (2007). Briefly, a 0.7% agarose block was embedded in a paraffin block. The TMA MASTER (3DHITECH) device was used to pre-perforate the 1 mm core in the acceptor block, and the 1 mm core was removed from the cell line control donor block and the TMA acceptor was used. Placed within the block to create a cell line control TMA, the TMA block was then pressed face down on a glass slide and the block was allowed to melt so that the paraffin melted and the core within the block was fully fused. Place on a hot plate at 15 ° C for 15 minutes Accordingly, the slide having aligning the core. Block was cooled, removed TMA block from the slide, trimmed, cut serial sections 5μm from TMA block.

B. Preparation of Mini-TMA To make a Mini-TMA, we used a formalin-fixed paraffin-embedded (FFPE) prostate tumor sample block in an annotated cohort of patients who underwent radical prostatectomy and were determined for Gleason score. used. This cohort consisted of about 40 indolent tumors (Gleason ≦ 3 + 3) and about 40 aggressive tumors (Gleason ≧ 4 + 3).

  TMA blocks were made using a modified agarose block procedure (Yan et al., Supra). Briefly, a 0.7% agarose block was embedded in a paraffin block and used as a TMA acceptor block. Using a TMA MASTER (3DHISTTECH) device, the acceptor block was pre-drilled for a 1 mm core. A 1 mm core was removed from about 80 cohort donor blocks and placed in a TMA acceptor block to create a mini TMA. Cell line controls were interspersed with the cohort samples and served as controls for intra-slide or inter-core staining reproducibility, inter-slide reproducibility, and daily staining reproducibility. The TMA block is then pressed face down on a glass slide and the core is removed by placing the block on a 65 ° C. hotplate for 15 minutes so that the paraffin can melt and the core in the block can fully fuse. Aligned. The slide with the block was cooled, the TMA block was removed from the slide, trimmed, and 5 μm serial sections were cut from the TMA block.

C. Preparation of High Measure Gleason TMA and Low Measure Gleason TMA An annotated cohort of formalin-fixed paraffin-embedded (FFPE) prostate tumor sample blocks from patients undergoing radical prostatectomy were obtained from Folio Biosciences (Powell, OH) .

  A series of 5 μm sections were cut from each FFPE block and used for tissue quality control processing and subsequent Gleason score annotation. Immunofluorescent staining was performed on some sections to see if the quality of the tissue was suitable for further testing and to ensure that the tissue contained sufficient tumor area for further testing. Briefly, this control FFPE section was processed for immunofluorescent staining, anti-phosphoSTAT3-T705 rabbit monoclonal antibody (mAb), anti-STAT3 mouse mAb, Alexa 488-conjugated anti-cytokeratin 8 mouse mAb, Alexa 488- Stained with conjugated anti-cytokeratin 18 mouse mAb, Alexa 555-conjugated anti-cytokeratin 5 mAb and Alexa 555-conjugated anti-TRIM mAb (see Table 1). The slides were visually inspected under a fluorescent microscope (Vectra System, PerkinElmer) for staining with each antibody. Based on staining intensity and autofluorescence, sections and their corresponding FFPE blocks were classified into four categories indicating tissue quality as shown in Table 2. The tumor area was defined as prostate epithelial structure without basal cell markers. Anti-cytokeratin 8 and anti-cytokeratin 18 mAb were used to show epithelial specific staining. Anti-cytokeratin 5 and anti-TRIM29 mAb were used to show basal cell staining. Only FFPE blocks that contained a sufficient amount of tumor area and included in the top two quality categories were used for further studies.

  The final 5 μm section cut from each FFPE block was stained with hematoxylin and eosin (H & E) and scanned using the Aperio XT system (Aperio, Vista, Calif.). Scanned images were stored in the SPECTRUM database (Aperio, Vista, CA). Remotely view images of H & E stained sections, and Gleason scores in a blinded manner in US clinical pathologists certified trials at Brigham and Women's Hospital (Boston, Mass.) And Johns Hopkins University (Baltimore, MD) An anatomical pathologist annotated with ImageScope software (Aperio, Vista, CA). The pathologist put annotated circles corresponding to a 1 mm core in the four regions with the highest Gleason score pattern and the two lowest values in the image of each section (see, for example, FIG. 1). One Gleason intercept with the highest value and one Gleason intercept with the lowest value were selected for inclusion in the high measurement Gleason TMA and low measurement Gleason TMA, respectively. If the tumor was relatively uniform, the high and low values were nearly identical.

  TMA blocks were made using a modified agarose block procedure (Yan et al., Supra). Briefly, a 0.7% agarose block was embedded in a paraffin block and used as a TMA acceptor block. Using a TMA Master (3D Histech) device, the acceptor block was pre-drilled for a 1 mm core. The 1 mm core was removed from the cell line control donor block (above) and placed in three separate regions of the acceptor block: the upper, middle and bottom portions. With this arrangement, cell line controls could serve as controls for intra-slide or inter-core staining reproducibility, inter-slide reproducibility, and daily staining reproducibility. One important feature of the cell line control was the consistency between remote sections of the TMA block. While tissue samples change as the core is cut into sections, the cell line control is a uniform mixture of cells throughout the core depth and does not change.

  The FFPE block of prostate tumor samples that passed quality control was selected as the patient sample donor block. The cores of these donor blocks were excised as per pathologist's annotation in the area corresponding to the selected high and low reading Gleason sections. The order of patient sample placement in the acceptor block was random. Duplicate cores are taken from each donor block (ie, one high measurement Gleason core and one low measurement Gleason core) and placed in one of two separate acceptor blocks The second core was placed at a position randomly extracted with respect to the position of the first core. In other words, the high reading Gleason TMA was randomly sampled separately from the low reading Gleason TMA. Thus, the two duplicate TMA blocks obtained were identical in that they were patient sample compositions, but their positions were randomized. The TMA block is then pressed face down on a glass slide and the core is removed by placing the block on a 65 ° C. hotplate for 15 minutes so that the paraffin can melt and the core in the block can fully fuse. Aligned. The slide with the block was cooled, the TMA block was removed from the slide, trimmed, and 5 μm serial sections were cut from the TMA block. The pathologist then annotated each core obtained from the prostate tumor sample and obtained a Gleason score measurement (only the isolated core, apart from the previously obtained “real” Gleason score of the whole tumor) On the basis of the). For example, a core selected from aggressive tumors and placed on LTMA is annotated as having a “measured” Gleason score of 3 + 3, even if the “actual” surgical Gleason score of the tumor is greater than 4 + 3. Can be.

Example 2: Engine for biomarker selection We have developed a biomarker selection and validation engine that can be used to identify biomarkers of any disease or condition (Figure 2). The engine has four main stages: a biological stage, a technical stage, a performance stage and a verification stage.

  Biological stage, initial biomarker candidate list is compiled from publicly available data for the disease of interest, eg from scientific literature, databases and conference presentations. The biomarker list is then prioritized based on biological relevance, in silico analysis, Human Protein Atlas review and the availability of the monoclonal antibodies required. The study of biological relevance is based on the mechanism of action in cells, particularly in cells of the disease. In silico analysis is based on previously known gene amplifications, deletions and mutations as well as univariate performance or progression correlations between such genetic modifications and diseases. Human Protein Atlas provides protein expression levels in various tissues in disease states. Biomarkers are ranked based on whether they are expressed at a range of expression levels in healthy and disease states.

  In the technical stage, commercially available antibodies are obtained from suppliers and tested for the ability to detect markers from clinical samples. First, antibodies are tested in immunohistochemical assays using classic 3,3'-diaminobenzidine (DAB) staining to assess the specificity and intensity of staining. Because DAB is more sensitive than immunofluorescence staining, it is important to identify markers that are detected by DAB with sufficient intensity to be detected by immunofluorescence. Antibodies and markers that meet DAB criteria are then evaluated by immunofluorescence (IF) to examine specificity, signal intensity, and kinetics (ie, range of expression). Advance antibodies and markers that meet IF criteria to the performance stage.

  In the performance stage, antibodies are tested with mini-TMA. Performance is evaluated for univariate correlation between expression and disease state. Antibodies and markers that exhibit a univariate correlation between expression and disease state are then evaluated for both univariate correlation and performance in combination with other markers in a larger TMA cohort. The major biomarker combinations are then validated using a clinical validation cohort.

Example 3 Selection of Prostate Cancer Biomarkers Using the biomarker selection engine described in Example 2, biomarkers for the identification of indolent and aggressive prostate cancer were tested as shown in FIG. Selected.

A. Biological stage Review of prostate cancer literature to identify markers associated with prostate cancer in mouse models, Gleason classification specific expression, correlation with progression, biology in prostate cancer Were compiled based on specific roles and / or known prostate cancer markers. Since some of the identified markers were part of one or more signaling pathways, other members of the signaling pathway were included in the initial candidate list. In total, 160 potential markers were included in the initial candidate target list.

  The initial target list was prioritized based on biological relevance, in silico analysis, Human Protein Atlas (available at www.proteinatlas.org/) and antibody availability. In assessing biological relevance, oncogenes and tumor suppressor genes were not considered as important in terms of prognosis, probably due to their low association with tumor grade. Similarly, priorities of genes identified in the in silico analysis as correlated with univariate performance and progression were also determined. However, the correlation between gene and protein expression is insufficient in prostate cancer. Therefore, most prioritization of prostate cancer markers is based on Human Protein Atlas showing the spatial distribution of proteins in 46 different normal human tissues and 20 different cancer types and 47 different human cell lines. did. In particular, the priority of proteins whose expression levels are different in different tumors was determined. This is because its expression level may be more closely correlated with tumor progression classification. After this analysis, a list of about 120 prioritized candidates was advanced to the technical verification stage.

B. Technical stage Antibodies against 120 prioritized candidates were obtained from commercial sources and verified by immunohistochemistry. Sections of various benign and cancerous prostate FFPE tissue samples were stained with candidate antibodies using a universal polymer DAB detection kit (ThermoFisher) using standard DAB protocols. Nearly half of the test antibodies showed a specific staining pattern at high intensity and were therefore selected for evaluation by immunofluorescence.

  Sections of various benign and cancerous prostate FFPE tissue samples were stained with a candidate antibody and a control cell line TMA using the immunofluorescence protocol described below. Antibodies showing a specific staining pattern were selected for further testing.

  Prostate cancer typically expresses epithelial markers such as cytokeratin 8 (CK8 or KRT8) and cytokeratin 18 (CK18 or KRT18), but a prostate basal marker such as cytokeratin 5 (CK5 or KRT5) Is a non-expressing cancer. The inventors have also surprisingly found that TRIM29, a tumor marker for some other cancers, is a basal marker rather than a tumor marker in prostate tissue; Can be used in masks. We evaluate tumor sections with a mixture of anti-CK5, anti-CK8, anti-CK18 and anti-TRIM29 antibodies, where the prostate tumor area is delineated by anti-CK8 and anti-CK18 antibodies, CK5 and anti-TRIM29 antibodies are defined as prostate tissue regions that are not bounded.

  5 μm sections were cut from cell line control TMA blocks and placed on HISTOGRIP (Life Technologies) coated slides. Slides were baked at 65 ° C. for 30 minutes, deparaffinized by continuous incubation in xylene, and rehydrated in a series of graded alcohols. Antigen activation was carried out at 95 ° C. for 40 minutes in a 0.05% anhydrous cytocolic acid solution having a pH of 7.4.

  Immunofluorescence staining was performed using LabVision Autostainer, all incubations were at room temperature, all washes were performed with TBS-T (TBS + 0.05% Tween 20), and all antibodies were diluted with TBS-T + 0.1% BSA solution. First, slides were blocked with Biotin Block (Life Technologies) solution A for 20 minutes, washed, then blocked with solution B for 20 minutes, washed, then blocked with Bacground Snapper (Biocare Medical) for 20 minutes and washed again. Either mouse or rabbit primary antibody was applied and incubated for 1 hour. In some cases, a mouse primary antibody against the first biomarker and a rabbit primary antibody against the second biomarker were applied to the slide and incubated for 1 hour.

  After extensive washing, either biotin-conjugated anti-mouse IgG or FITC-conjugated anti-rabbit IgG was applied for 45 minutes. When two biomarkers were detected on the same slide, both biotin-conjugated anti-mouse IgG and FITC-conjugated anti-rabbit IgG were applied for 45 minutes. After thorough washing, streptavidin-Alexa 633, anti-FITC mAb-Alexa 568 and Tumor Mask cocktail (anti-cytokeratin 8 mAb Alexa 488, anti-cytokeratin 18 mAb Alexa 488, anti-cytokeratin 5 mAb Alex5 55 TRA55 55) A mixture of Alexa fluorophore-conjugate reagents consisting of

To enable automated image analysis of prostate cancer tumor tissue, we used a combination of antibodies to prostate epithelial and basal markers (Tumor Mask) and object recognition based on Definiens Developer XD (described below). . Tumor area was defined as prostate epithelial structure without basal marker. Epithelium-specific staining was obtained using Alexa 488-conjugated anti-cytokeratin 8 and anti-cytokeratin 18 specific mouse mAb cocktail. Basal cell staining was based on a cocktail of Alexa 555-conjugated anti-cytokeratin 5 and anti-TRIM29 specific mAb. Slides were incubated with these Alexa fluorophore-conjugate reagents for 1 hour. After thorough washing, a DAPI solution (100 ng / ml DAPI in TBS-T) was applied for 3 minutes. After several washes, the slides were placed in Prolong Gold antifade reagent (Life Technologies). Slides were allowed to “set” overnight in the dark at −20 ° C. and stored for long periods in the dark at −20 ° C. to minimize discoloration. The amount of immunofluorescence for each marker was evaluated. For example, FIG. 4 shows quantitative immunofluorescence of two different markers (FUS and DERL1) in two different sections (sections 27 and 41) of the control cell line TMA. The amount of immunofluorescence detected in each cell line in section 27 is shown on the x-axis, while the amount of immunofluorescence detected in each cell line in section 41 is shown on the y-axis. The linear relationship of the amount of immunofluorescence in these two cell lines and the high R ² value indicate the reproducibility of the quantitative immunofluorescence assay between experiments.

  Next, we tested the range of marker expression. For optimal dilution of the marker antibody in our staining assay and for reproducibility, we created a “Titration” TMA. We selected 40 FFPE blocks of prostate cancer samples with a series of Gleason scores. A “titration” TMA was then made using duplicate cores of each donor sample using the modified agarose block procedure described above. Immunofluorescence staining with a single marker and tumor region recognition anti-cytokeratin 8, anti-cytokeratin 18, anti-cytokeratin 5 and anti-TRIM29 antibody was performed. As discussed above, for detection of mouse monoclonal candidate antibodies, we used anti-mouse-biotin secondary and streptavidin-Alexa 633 tertiary antibody. Anti-rabbit-FITC secondary antibody and anti-FITC mAb-Alexa 568 tertiary antibody were used as the rabbit monoclonal candidate antibodies. Images were acquired with the Vectra system described below, and marker expression was quantified using Inform 1.3 software. Based on the specificity of the marker, the signal intensity and the dynamic range of the marker, 62 verified candidates were advanced to the performance stage.

C. Performance Stage Minicohort Screening The 62 candidates verified were tested in the mini TMA produced as described above in Example 1. Quantitative immunofluorescence assays were performed using the mouse and rabbit primary antibodies described above at the technical stage. 62 biomarkers were quantified to examine the difference in expression levels between about 40 indolent tumor samples and about 40 aggressive tumor samples. Of the 62 markers, 33 showed univariate performance with respect to correlation with indolent or aggressive tumor status.

  The 33 univariate performance markers were tested in the extended biopsy simulation test using the high measurement Gleason TMA and the low measurement Gleason TMA (HLTMA). Because the Gleason score measured for each core with high and low TMA may differ from the actual Gleason score of the tumor from which the core was obtained (based on the entire tumor removed surgically), It is possible to identify biomarkers that predict true Gleason score and thus aggressiveness, independent of the location of. In other words, we expected to identify biomarkers that could minimize sampling bias caused by intra-tumor heterogeneity. For example, each of the indolent, intermediate and aggressive tumors were presented in a low reading TMA (see, eg, FIG. 5 for a summary of the actual Gleason score of the low reading TMA core).

  For this 33 biomarkers with univariate performance in mini TMA experiments, quantitative immunofluorescence assays were performed using the mouse and rabbit primary antibodies described above at the technical stage. Two of the markers were discarded due to technical difficulties during HLTMA immunostaining and detection. Therefore, 31 types of biomarker data with univariate performance were acquired and analyzed in the mini TMA experiment.

D. Image Acquisition Two Vectra Intelligent Slide Analysis Systems (PerkinElmer) were used for quantitative multiplex immunofluorescence (QMIF) image acquisition. The TMA acquisition protocol was performed according to the manufacturer's instructions (with minor modifications). The same exposure time was used for all slides. To minimize differences between TMA, TMA slides stained with the same antibody combination were processed on the same Vectra microscope.

  DAPI, FITC, TRITC and Cy5 long pass emission filter cubes were obtained from Semirock. TRITC and Cy5 filter cubes were optimized to allow maximal spectral separation between Alexa 555, Alexa 568 and Alexa 633 dyes. DAPI, FITC, TRITC and Cy5 long pass emission filter cubes were obtained from Semirock. TRITC and Cy5 filter cubes were optimized to allow maximal spectral separation between Alexa 555, Alexa 568 and Alexa 633 dyes.

  Acquisition of the DAPI band was performed in 20 nm steps. Acquisition of FITC, TRITC and Cy5 bands was performed in 10 nm steps. Two 20 × image cubes per core were acquired with a sequential collection of DAPI, FITC, TRITC and Cy5 band images. Spectral libraries were generated according to the manufacturer's instructions and the image cubes were unmixed into floating TIFF files with individual fluorophore and autofluorescence signals using Inform 1.4 software (PerkinElmer). Two channels were created for autofluorescence (one for general tissue autofluorescence and one for red blood cells) and light granules were scattered throughout the prostate tissue. After unmixing the images, the TIFF fileset was further analyzed with Definiens Developer software. For analysis of smaller “titration” TMA data, the image cube was unmixed using Inform 1.3 software (PerkinElmer) to quantify marker expression.

  To examine if there is a device-to-device difference (if any) between the two Vectra Intelligent Slide Analysis Systems (PerkinElmer), we performed CTMA in parallel on these two devices, Alexa-568, Alexa- The detection of 633 and Alexa-647 was analyzed. As shown in FIG. 6, the two systems differed by less than 2% for Alexa-647 detection and about 7% for Alexa-568 detection. However, Alexa-633 detection differed by about 20% between the two devices. Using this data, we were able to establish an interdevice conversion count for each channel.

E. Image Analysis A fully automated image analysis algorithm for tumor identification and biomarker quantification was created using Defienens Developer XD ^™ (Definiens, Inc., Parsippany, NJ) (see, eg, FIG. 7). Two 20 × 1.0 mm image fields were obtained for each tissue microarray (TMA) core. The Vectra multispectral image file was first converted to a multi-layer TIFF format using inForm, and then converted to a single layer TIFF file using BioFormats. Single layer TIFF was imported into the Definiens workspace using a customized import algorithm. For each TMA core, both image field TIFF files were loaded as “maps” into a single “scene” as per manufacturer's instructions.

Built-in automatic adaptive binarization was used to define a fluorescence cutoff for tissue segmentation with each individual tissue sample in our image analysis algorithm. Cell line controls were automatically identified based on pre-defined core positions. Tissue samples were segmented using fluorescent epithelial and basal cell markers with DAPI for classification into epithelial cells, basal cells and stroma, and further partitioned into cytoplasm and nucleus. Cell line controls were segmented using autofluorescent channels. Fields with artifact staining, insufficient epithelial tissue and defocus were removed by a strict multi-parameter quality control algorithm (see, eg, FIG. 8). Individual gland regions in the tissue sample are further classified as malignant or benign based on relevant features between basal cells and adjacent epithelial structures in combination with object-related features such as gland thickness (eg, FIG. 9 reference).

  The intensity of epithelial markers and DAPI were quantified as quality control measures in malignant and non-malignant epithelial areas. The biomarker value was measured independently in the cytoplasm, nucleus or whole cell of the malignant tissue based on the predetermined subcellular localization (for example, see FIG. 10). The average biomarker pixel intensity in the malignant compartment was averaged in a map with acceptable quality parameters to obtain a single value for each tissue sample and cell line control core.

  Data obtained from the Definiens analysis was exported for bioinformatics analysis or Clinical Lab Improvement Ambient Information System (LIS) analysis.

F. Data analysis Mean biomarker values obtained for 31 biomarkers with univariate performance in a mini TMA experiment were examined for their correlation with tumor aggressiveness and lethality. As discussed above, indolent, intermediate and aggressive tumors were each presented in both high and low measurements TMA (see, eg, Figure 5 for details of each category for low measurements TMA) ) In an aggressive test, we determined that the biomarker expression was aggressive and a set of four different samples: (1) a core with all Gleason score measurements ≦ 3 + 3; (2) all Gleason score measurements ≦ 3 + 4 (In this case, cores with an intermediate surgical Gleason score are excluded); (3) cores with all Gleason score measurements ≦ 3 + 4 (in this case, cores with an intermediate surgical Gleason score are aggressive) And (4) cores with all Gleason score measurements ≦ 3 + 4 (in this case, cores with an intermediate surgical Gleason score were correlated in counting as indolent (FIG. 11).) In this study, we used biomarker expression for lethal invasion and two different sample sets: (1) all And (2) correlated in all cores (Figure 10) Biomarker values were correlated on a univariate basis using T-test, Wilcoxon test and permutation test. Of the 31 biomarkers, 17 biomarkers showed univariate performance in both the measurement of aggressive outcome and the measurement of lethal outcome (FIG. 11).

Next, we whether the combination of the biomarkers are correlated with tumor aggressiveness, two different approaches: (1) univariate performance in both the measurement of the measurement and lethal outcomes aggressive outcome And evaluated using an unbiased analysis of all 31 biomarker combinations tested in the HLTMA analysis (FIG. 12). A combination of 3-10 biomarkers selected from 17 univariate performance biomarkers was analyzed, and a combination of 3-5 biomarkers from a set of 31 biomarkers was analyzed. For each marker combination, 500 training case sets were created by bootstrap (ie, restoration extraction) to obtain related test case sets. The model was derived by logistic regression on the training case set and tested in the related test case set. Training and testing C statistics (ie, area under the curve) and training Akaike Information Criterion (AIC) were obtained each time. Median and 95% confidence intervals were obtained for all three statistics. Table 3 shows the top rank models of tumor aggressiveness in pre-selected combinations for univariate performance with each analysis method.

  As expected, when data was stored based on training data or AIC, the correlation between various combinations and aggressiveness increased as the size of the combinations increased. In other words, the 10-member combination has a higher predictability of invasion than the 9-member combination. This is predicted because each additional member in the combination adds additional degrees of freedom to the training analysis. However, when data was stored based on test data, the 7 member or 6 member combination was more correlated than the 8, 9 and 10 member combinations. This is because it becomes more difficult to generalize the test data as the data is trained with a greater degree of freedom. Thus, in some cases a combination of 6 or 7 biomarkers may be more useful in predicting tumor aggressiveness in clinical assays. For each AIC and test data, the frequency of which each biomarker appears in the top combination was examined. Figure 13 for the top 1% top biomarkers of the 3-10 member combination stored in the AIC; Figure 14 for the top 5% top biomarkers of the 3-10 member combination stored in the AIC; and AIC See FIG. 15 for the top 1% and top 5% top biomarkers of the 7 member combinations stored in the test C statistic data. The top 5% of the 7 member combinations stored in the test data are disclosed in US Provisional Patent Application No. 61 / 792,003 (filed on Mar. 15, 2013, the entire contents of which are incorporated herein by reference) Table 6 is shown.

Table 4 shows the top rank models of tumor aggressiveness in combinations that were not pre-selected for univariate performance with each analysis method. For each AIC and test data, the frequency of which each biomarker appears in the top combination was examined. See FIG. 16 for the frequency of which biomarkers appear in the top 1% of the five member combinations stored in the AIC and test data. See FIG. 17 for the frequency of which biomarkers appear in the top 5% of the five member combinations stored in the AIC and test data. The flat tails of FIGS. 16 and 17 suggest that many of these biomarkers are interchangeable and little performance addition can be shown.

  When all the above data are combined, a core set of 7 markers (ie, ACTN1, FUS, SMAD2, HOXB13, DERL1, RpS6 and CoX6C) that appear consistently in all selection approaches for prostate tumor aggressiveness is identified (See markers with 100% or 75% in the rightmost column of FIG. 18). A secondary set of 7 markers of prostate tumor aggressiveness (ie, YBX1, SMAD4, VDAC1, DCC, CUL2, PLAG1 and PDSS2) can also be easily identified (50% in the rightmost column of FIG. 18). (See Marker with).

We also determined whether the combination of biomarkers correlates with lethal outcomes, as well as the 17 biomarkers that showed univariate performance in both aggressive and lethal outcome measures. The combination was used for evaluation. Combinations of 3-10 biomarkers selected from 17 univariate performance biomarkers were analyzed by logistic regression. A training / test cohort using bootstrap method (ie, restoration extraction) and multiple cross-validations was analyzed by C statistic, AIC and 95% confidence interval. Table 5 shows the top rank models of lethal outcomes for pre-selected combinations for univariate performance with each analysis method.

  Similar to the tumor aggressiveness model described above, a combination of 6 or 7 biomarkers may be most useful in predicting lethal outcome in clinical assays. For each AIC and test data, the frequency of which each biomarker appears in the top combination was examined for lethal outcome analysis. 19 for the top 1% top biomarkers of 3-10 member combinations stored in AIC; FIG. 20 for the top 5% top biomarkers of 3-10 member combinations stored in AIC; and AIC See FIG. 21 for the top 1% and top 5% top biomarkers of the 7 member combinations stored in the test data. Interestingly, comparing FIGS. 15 and 21, eight biomarkers appear in the top 12 biomarkers in both tumor invasion and lethal outcome. It is therefore possible to select a set of biomarkers that partially overlap in the likelihood of predicting tumor invasion and lethal outcome (FIG. 22).

Example 4: Clinical validation of biomarker combinations Using the top biomarker combinations identified in Example 3 above, we evaluated clinical tumor samples for tumor invasion and the invention described above An assay was designed to validate the results of these models. The image acquisition hardware can detect up to six different fluorescent channels. Thus, up to three biomarkers (or prognostic determinants “PD”) can be detected along with two tumor mask signals and nuclear staining (eg, DAPI), ie, triple staining. For example, FIG. 23 shows three biomarkers from a single slide (HSD17B4, FUS and LATS2), two tumor mask signals (CK8 + CK18-Alexa 488 and CK5 + TRIM29-Alexa 555) and six of nuclear staining (DAPI). The detection of different fluorescent signals is shown. For example, the first channel can be used to detect a first biomarker (eg, PD1). The primary antibody of this marker is conjugated to a FITC molecule and can be detected by anti-FITC-Alexa-568. The second channel can be used to detect a second biomarker (eg, PD2), the primary antibody of which is conjugated to biotin conjugated anti-rabbit _Fab and Alexa-633. Rabbit antibodies that can be detected using streptavidin. The third channel, the third biomarker (e.g., PD3) can be used to detect the primary antibody of this marker, and anti-mouse _{F ab} conjugated to horseradish peroxidase (HRP) Alexa- A mouse antibody that can be detected with anti-HRP conjugated to 647. The fourth channel can be used to detect cancer epithelial markers such as cytokeratin 8 (CK8 or KRT8) and cytokeratin 18 (CK18 or KRT18). For example, a combination of anti-CK8-Alexa-488 and anti-CK18-Alexa-488 can be used to define the tumor area of the sample. The fifth channel can be used to detect basal epithelial markers such as cytokeratin 5 (CK5 or KRT5) and TRIM29. For example, a combination of anti-CK5-Alexa-555 and anti-TRIM29-Alexa-555 can be used to define a non-tumor region of a sample. The sixth channel can be used to detect cell structure, for example to detect DAPI stained nuclei. Core Set of 7 Markers From a secondary set of 7 markers, we identified 12 commercially available antibodies against these markers suitable for triple staining on quadruplicate slides ( (See FIG. 24).

  In order to confirm that staining with multiple types of antibodies in triple staining cannot affect the detection of the antibodies, we have determined that bios in the presence of the assay alone or another biomarker (PD2) antibody. The signal from the antibody against the marker (PD1) was compared in the background of the tumor mask marker. As shown in FIG. 25, the addition of an antibody to the second biomarker has minimal impact on the detection of the first marker. Using this analysis, we confirmed that the marker combinations shown in FIG. 24 could be used with minimal hindrance in the detection of each biomarker.

  Next, we obtained two prostate cancer tumor sample cohorts: a cohort of 350 tumors from the Cleveland Clinic and a cohort of 180 tumors from the Harvard School of Public Health. We isolated five 5 μm serial sections from each tumor sample in the cohort. Four of the sections were used for biomarker detection as shown in FIG. The fifth section was used to examine the quality of the tumor sample by assessing autofluorescence. Two channels were evaluated for autofluorescence (one for general tissue autofluorescence (AFL) and one for red blood cells) and scattered throughout the light granule (BAFL) prostate tissue. For example, see FIG.

  Specifically, 5 μm sections were cut from paraffin-embedded tumor sample blocks and placed on Histogrip (Life Technologies) coated slides. Slides were baked at 65 ° C. for 30 minutes, deparaffinized by continuous incubation in xylene, and rehydrated in a series of graded alcohols. Antigen activation was carried out at 95 ° C. for 40 minutes in a 0.05% anhydrous cytosolanic acid solution at pH 7.4.

  Immunofluorescence staining was performed using LabVision Autostainer, all incubations were at room temperature, all washes were performed with TBS-T (TBS + 0.05% Tween 20), and all antibodies were diluted with TBS-T + 0.1% BSA solution. First, slides were blocked with Biotin Block (Life Technologies) solution A for 20 minutes, washed, then blocked with solution B for 20 minutes, washed, then blocked with Bacground Snapper (Biocare Medical) for 20 minutes and washed again. A mixture of FITC-conjugated mouse primary antibody and rabbit primary antibody (see Figure 23) was applied and incubated for 1 hour.

  After extensive washing, a mixture of biotin-conjugated anti-rabbit IgG and HRP-conjugated anti-rabbit IgG was applied for 45 minutes. After thorough washing, streptavidin-Alexa 633, anti-FITC mAb-Alexa 568, anti-HRP mAb-Alexa 647 and Tumor Mask cocktail (anti-cytokeratin 8 mAb Alexa 488, anti-cytokeratin 18 mAb Alexa 488, anti-cytokera 488 A mixture of Alexa fluorophore-conjugate reagents consisting of Alexa 555, anti-TRIM29 mAb Alexa555) was applied. As described above, the inventors have used an antibody to prostate epithelial marker and basal marker (Tumor Mask) and object recognition based on Definiens Developer XD to enable automated image analysis of prostate cancer tumor tissue did. The tumor area was defined as prostate epithelial structure without basal markers. Epithelium-specific staining was obtained using a cocktail of Alexa 488-conjugated anti-cytokeratin 8 and anti-cytokeratin 18 specific mouse mAbs. Basal cell staining was based on a cocktail of Alexa 555-conjugated anti-cytokeratin 5 and anti-TRIM29 specific mAb. Slides were incubated with these Alexa fluorophore-conjugate reagents for 1 hour. After thorough washing, a DAPI solution (100 ng / ml DAPI in TBS-T) was applied for 3 minutes. After several washes, the slides were placed in Prolong Gold antifade reagent (Life Technologies). Slides were allowed to “set” overnight in the dark at −20 ° C. and stored for long periods in the dark at −20 ° C. to minimize discoloration. Images were acquired and analyzed as described in Example 3.

Example 5: Development of an automated quantitative multiplex proteomics in situ imaging platform and application overview in predicting prostate cancer lethal outcomes Gene-based approach to cancer prognosis, promising early intervention for high-risk patients and overtreatment for low-risk patients Significant progress has been made in avoiding this. Despite protein level perturbations and post-translational modifications being more directly associated with phenotypes, little progress has been made in proteomic approaches. Most gene expression system platforms currently in use require tissue lysis, resulting in loss of structural molecular information, and thus do not have the disadvantages of tumor heterogeneity and morphological features. Here we present an automated integrated multiplex proteomics in situ imaging platform that measures protein biomarker levels and activity state quantitatively in a defined intact tissue region where the biomarker of interest exerts its phenotype. As a proof of concept, it was confirmed that the lethal outcomes of human prostate cancer can be predicted by four previously reported prognostic markers PTEN, SMAD4, CCND1 and SPP1. Furthermore, the mechanistic power of protein expression was shown by showing that PTEN can be replaced by two PI3K pathway regulatory protein activities. In summary, the platform can reproducibly and simultaneously quantify and evaluate multiple functional activities of oncogenes and tumor suppressor genes in intact tissues. The platform is widely applicable and well suited for the prognosis of early-stage disease classifications in which important signaling protein levels and activities are disrupted.

INTRODUCTION Although tests for recurrence-verifying gene mutations have significant prognostic and predictive aspects, ^1-5 , such mutations are relatively rare in early stage cancers. Multivariate gene-based testing requires homogenized tissue with varying proportions of tumor and benign tissue, resulting in low accuracy biomarker measurements ^6,7 . In these types of tests, the phenotype must be inferred from the gene pattern and the mutation pattern. In contrast, direct in situ measurements of protein levels and post-translational modifications should more directly reflect the status of oncogene signaling pathways. It is therefore reasonable to predict that a protein-based approach will be beneficial for prognosis.

Other problems also complicate prognostic testing. In prostate cancer, tumor heterogeneity is significant and sampling errors can contribute to inaccurate predictions. Prognosis in this multifocal disease is difficult even by pathologists who do not follow the Gleason classification and the tumor progression classification. To address these shortcomings, we have automated quantification for intact tissue with integrated morphological object recognition and molecular biomarker measurements of defined related tissue regions at the individual slide level. Multiplex proteomic imaging platform was developed. This system was used to predict the lethal outcome with radical prostatectomy from resection tissue previously reported four markers PTEN, SMAD4, CCND1 and SPP1 ^8. Importantly, here also a quantitative measurement of the activity state of the protein reflecting the PI3K / AKT and MAPK signaling states is a highly validated prognostic marker that itself modulates PI3K / AKT pathway signaling. It is also shown that some PTEN ^9-13 can be substituted. Taking these data together, new lethal outcome predictors were identified: SMAD4, CCND1, SPP1, phospho-PRAS40 (pPRAS40) -T246 and phospho-ribosome S6 (pS6) -S235 / 236.

Materials and Methods Reagents and Antibodies All antibodies and reagents used in this study were obtained from commercial sources listed in Table 7. Anti-FITC MAb-Alexa 568, Anti-CK8-Alexa 488, Anti-CK18-Alexa 488, Anti-CK5-Alexa 555 and Anti-Trim29-Alexa 555 have Alexa dyes and their own appropriate protein conjugation kits (Manufacturer's instructions (LifeTechnologies, Grand Ills). , NY).

Formalin fixed paraffin embedded (FFPE) prostate cancer tissue block acquisition, processing and quality control

  We obtained a cohort of FFPE human prostate cancer tissue blocks with clinical annotations and long-term patient outcome information from Folio Biosciences (Powell, OH). Samples were collected with appropriate IRB approval and all patient records were anonymized. We have several FFPE human prostate cancer tissue blocks from other commercial sources (BioOptions, Brea, CA; Curline, So. San Francisco, CA; ILSBio, Chestertown, MD; OriGene, Rockville, MD). In addition, we verified the staining intensity of multiplex staining formats for individual antibodies and combinations, developed a quality control procedure, evaluated in-vivo reproducibility tests, and confirmed the specificity of staining in prostate tumor tissue.

Ten to twelve sections (5 μm cut) were made from each FFPE block. The last section was stained with hematoxylin and eosin (H & E) and scanned with the Aperio (Vista, CA) XT system. H & E stained images were stored in the Spectrum database (Aperio, Vista, Calif.) For remote observation and focused Gleason annotation in a blinded fashion by a professional certified anatomical pathologist using ImageScope software (Aperio, Vista, Calif.). The four regions with the highest Gleason pattern of each prostatectomy sample and the two regions with the lowest were marked with an annotated circle corresponding to a 1 mm core using the current criteria ¹⁴ .

Tissue Quality Control Procedure 5 μm sections from each FFPE block were stained with anti-phosphoSTAT3T705 rabbit monoclonal antibody (mAb), anti-STAT3 mouse mAb and region of interest marker as described below. Slides were examined under a fluorescence microscope. Based on staining intensity and autofluorescence, tissues were qualitatively classified into four categories as shown in Table 8 and FIG. 27E. FFPE blocks belonging to the top two quality categories were included in the test.

Definition:
High-bright fluorescent staining. Uniformly low for CK8 and 18-barely visible staining, partly, no background level-no staining observed

Cell Line Controls Cell lines selected for use as positive and negative controls were cultured under standard conditions, treated with drugs and inhibitors, and collected as indicated (Table 10). Cells were washed with PBS and fixed directly on the plate with 10% formalin for 5 minutes, then scraped and collected in PBS. The cells are then washed twice with phosphate buffered saline (PBS), resuspended in Histogel (Thermo Scientific, Waltham, Mass.) At 70 ° C., spun for 5 minutes (10,000 g), and concentrated. A cell-histogel pellet was formed. The pellet was embedded in paraffin to form a standard paraffin block and used as a donor block for the construction of a tumor microarray.

Generation of Tumor Microarray (TMA) Blocks TMA blocks were generated using the modified agarose block procedure ¹⁵ . Briefly, a 0.7% agarose block was embedded in paraffin and used as a TMA acceptor block. Two 1 mm diameter cores were drilled into the donor block from the region corresponding to the highest Gleason pattern according to the pathologist's annotation using a TMA Master (3D Histech, Budapest, Hungary) instrument. One of these cores is placed at a randomly sampled position in one acceptor block, while the other core's position in the second acceptor block is randomly sampled relative to the first core. did. This was repeated for 91, 170 and 157 annotated prostate tumor samples (Table 9) to form three pairs of TMA blocks (MPTMAF 1A and 1B, 2A and 2B, 3A and 3B), respectively. The resulting block of pairs was identical in that it was a patient sample composition, but was sampled randomly with respect to sample location. Cell line control cores were added to the upper, middle and bottom portions of these acceptor blocks. Once loaded, the TMA block was pressed face down on a glass slide at 65 ° C. for 15 minutes to allow fusion of the TMA core into the host paraffin. The paraffin block was then cut into 5 μm serial sections. Smaller test TMAs were made from commercially available FFPE prostate tumor cases with only limited (Gleason score) annotations. Using this TMA, PTEN values were compared using phosphomarkers before the main cohort test to confirm reproducibility. Reproducibility was shown by comparing individual marker signals in serial sections of test TMA (Table 9 and Figure 27F).

Slide Processing and Quantitative Multiplex Immunofluorescence (QMIF) Staining Protocol Five um thick TMA sections were cut and placed on Histogrip (Life Technologies, Grand Island, NY) coated slides. Slides were baked at 65 ° C. for 30 minutes, deparaffinized by continuous incubation in xylene, and rehydrated in a series of graded alcohols. Antigen activation was performed using a PT module (Thermo Scientific, Waltham, Mass.) In a 0.05% cytosolanic anhydride solution at 95 ° C. for 45 minutes. Autostainers 360 and 720 (Thermo Scientific, Waltham, Mass.) Were used for staining.

  The staining procedure included four incubation steps (with appropriate washing between steps) after two blocking steps. Blocking consisted of Snipper reagent (Biocare Medical, Concord, CA) after the biotin step. The first incubation step included anti-biomarker 1 mouse mAb and anti-biomarker 2 rabbit mAb. The second step includes anti-rabbit IgG Fab-FITC and anti-mouse IgG Fab-biotin followed by a third “visualization” step, which includes anti-FITC MAb-Alexa568, streptavidin-Alexa633. And fluorophore-conjugated region of interest antibodies (anti-CK8-Alexa488, anti-CK18-Alexa488, anti-CK5-Alexa555 and anti-Trim29-Alexa555) were included. Finally, sections were incubated with DAPI for nuclear staining (see FIG. 27B for an overview of the staining format). The slide was placed on Prolong Gold (Life Technologies, Grand Island, NY), and a cover glass was placed thereon. Slides were kept at −20 ° C. overnight, then imaged and stored for long periods. A full set of 6 MPTMAF slides were stained in one staining session so that the various antibody combinations included all biomarkers tested.

Antibody validation Testing by Western blotting before and after knockdown: To test the specificity of mAbs for PTEN, SMAD4 and CCND1, we used inducible shRNA knockdown of the protein marker of interest. Briefly, DU145 cells with inducible shRNA were generated by transducing naive DU145 cells with a virus carrying pTRIPZ (Thermo Scientific, Waltham, Mass.). Cells were stably selected for 1 week using 2 μg / ml puromycin. Cells were subsequently induced for 72 hours with either 0.1 μg / ml or 2 μg / ml doxycycline. Cells were trypsinized and processed for either RNA extraction or cell lysate production. The best shRNA for each protein marker was confirmed first by RT-PCR and then by Western blot. The antibody was considered specific if the molecular band size expected to be shRNA induced in the Western blot was reduced. In order to test mAbs against SPP1, we used cell lines with high or low SPP expression. In addition, lysates of these cell lines (as shown in FIG. 30) were also used for Western blotting. To test anti-phospho antibodies against members of the AKT signaling pathway, DU145 cells were deprived of serum overnight and treated with 10 μM PI3K inhibitor LY294002 for 2 hours and lysed. Lysates of cells treated with inhibitors were used as negative controls for Western blots; lysates of cells cultured in standard conditions were used as positive controls.

20 μg of cell lysate was run on a 4-15% Criterion TGX precast gel (Bio-Rad, Hercules, Calif.). Thereafter, the gel was transferred to a nitrocellulose membrane using iBlot (Life Technologies, Grand Island, NY). Primary antibody dilutions were used according to the recommendations in the product data sheet. The film was developed using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific, Waltham, Mass.). Images were acquired using the FluroChem Q system (Protein Simple, Santa Clara, CA). Images, AlphaView (Protein Simple, Santa Clara , CA) was treated with and ImageJ ^16.

Testing by immunohistochemistry before and after target knockdown: FFPE cell pellets of cell lines treated as described above were combined into TMA blocks. 5 μm sections were cut and dried at 60 ° C. for 1 hour, then deparaffinized in xylene (3 changes) and rehydrated in a series of decreasing ethanol washes. Slides were heated at 95 ° C. for 40 minutes in 0.05% cytosolanic anhydride (Sigma, Saint Louis, Mo.) for antigen activation. The slides were stained using Lab Vision ^™ UltraVision ^™ LP Detection System: HRP Polymer / DAB Plus Chromogen Kit (Thermo Scientific, Waltham, Mass.) As per manufacturer's instructions. Slides were scanned with the Aperio Scanscope AT Turbo system (Aperio, Vista, CA). Images were analyzed with Aperio ImageScope software (Aperio, Vista, CA).

Image Acquisition Two Vectra Intelligent Slide Analysis Systems (Perkin-Elmer, Waltham, Mass.) Were used for automatic image acquisition. The DAPI, FITC, TRITC and Cy5 long pass filter cubes were optimized to allow maximum spectral resolution and minimize cross interference between fluorophores. Vectra 2.0 and Nuance 2.0 software packages (Perkin Elmer, Waltham, Mass.) Were used for automatic image acquisition and development of the spectral library, respectively.

  The TMA acquisition protocol was performed in automatic mode, following the manufacturer's instructions (Perkin-Elmer, Waltham, Mass.). Two 20 × fields per core were imaged using a multispectral acquisition protocol including continuous exposure with DAPI, FITC, TRITC and Cy5 filters. In order to ensure the reproducibility of biomarker quantification, the light source intensity was calibrated with an X-Cite Optical Power Measurement System (Lumen Dynamics, Mississauga, ON, Canada) prior to image acquisition of each TMA slide. All slides containing the same antibody combination used the same exposure time. To minimize intra-experimental differences, TMA slides stained with the same antibody combination were imaged with the same Vectra microscope.

  Spectral profiles of each fluorescent dye as well as FFPE prostate tissue autofluorescence were generated. Interestingly, two types of autofluorescence were observed in FFPE prostate tissue. While typical autofluorescence signals were common to both benign and tumor tissues, typical “bright” autofluorescence was specific to bright granules, mostly present in epithelial cells of benign tissue. It was. A spectral library containing a combination of the two spectral profiles was used to separate or “unmix” individual dye signals from the autofluorescent background (FIGS. 27A and 27C).

Image Analysis We have developed an automated image analysis algorithm for tumor identification and biomarker quantification using Definiens Developer XD (Definiens AG, Munich, Germany). Two 20 × image fields were acquired for each 1.0 mm TMA core. The Vectra multispectral image file was first converted to a multi-layer TIFF format using inForm (PerkinElmer, Waltham, MA), and then the customized spectral library was converted to a single-layer TIFF file using BioFormats (OME ¹⁷ ). Import a single-layer TIFF file into the Definiens workspace using a customized import algorithm so that for each TMA core, both image view TIFF files are loaded as "maps" into a single "scene" and analyzed did.

  Automatic fit binarization was used to define a fluorescence intensity cutoff for tissue segmentation at each individual tissue sample. Tissue samples were segmented using DAPI with fluorescent epithelial and basal cell markers to allow classification as epithelial cells, basal cells and stroma, and further partitioned into cytoplasm and nucleus. The benign prostate gland contains basal and luminal cells, but the prostate cancer gland has no basal cells and a low luminal profile. Thus, individual glandular regions were classified as malignant or benign based on a combination of related features between basal cells and adjacent epithelial structures and object related features such as gland size (see FIG. 27D). Fields with artifact staining, insufficient epithelial tissue or defocused images were removed prior to scoring.

  The intensity of epithelial markers and DAPI was quantified as a regional quality control measure in benign and malignant epithelial areas. The intensity level of the biomarker was measured based on a predetermined subcellular localization criterion in all cytoplasm, nucleus or cancer cells. The average biomarker pixel intensity within the cancer compartment was averaged in a map with acceptable quality parameters to obtain a single value for each tissue sample and cell line control core.

Patient Cohort Composition FIG. 28A shows the FOLIO cohort composition used in this study, including a comparison with PHS cohort ⁸ .

Measurement of marker values Since each sample was made up of two cores, we created an aggregate score for each marker based on a correlation instruction. For markers that were positively correlated with lethality, we used the core with the highest value; for markers that were negatively correlated, we used the core with the lowest value. . For example, for the tumor suppressor SMAD4 present in all stained sections, we used three cores with the lowest core value.

Univariate analysis A univariate cox model was trained for each biomarker. Hazard ratios and log rank p-values are calculated for each marker, with positive correlation markers comprising the top 1/3 and lower 2/3 risk scores, and negative correlation markers comprising the lower 3 risk scores. The population consisting of a fraction and the population consisting of the top two-thirds were compared (FIGS. 28B and C).

Multivariate analysis We used multivariate analysis to examine the possibility of predicting lethal outcomes with marker sets. We used two model design approaches and two matrices. Specifically, 10,000 bootstrap training samples were created and both multivariate Cox models and logistic regression models were trained on each training sample. Each test was performed on a complement set. Concordance index (CI) and area under the curve (AUC) were used to estimate model performance. Kaplan-Meier curves were generated and the population with the lower 2/3 risk score was compared to the population with the upper 1/3 risk score. For all cohorts, receiver operating characteristic (ROC) curves were generated based on risk scores from a logistic regression model. The marker combinations tested in our model are: 1) PTEN, SMAD4, CCND1 and SPP1, and 2) SMAD4, CCND1, SPP1 and the following phosphomarkers: pS6, pPRAS40 and pS6 + pPRAS40 One. FIG. 31 outlines the multivariate analysis approach.

Results Platform Development Several technical requirements for the development of an automated multiplex proteomic imaging platform: 1) multiple types of markers in a defined area of interest (ie in tumor and surrounding benign tissue (contrast)) It was necessary to meet the possibility of quantification, 2) strict tissue quality control, 3) balanced multiplex assay staining format, and 4) experimental reproducibility.

To address the first, we have optimized long-pass DAPI, FITC, TRITC and Cy5 filter sets to have sufficient excitation and emission bandpass and to minimize interchannel interference. did. We further separated the biomarker signal from endogenous autofluorescence by spectral unmixing of the image (Figure 27A; ¹⁸ ). In order to quantitatively measure biomarkers only within the tumor epithelium, we needed to “tissue segmentation” to distinguish the tumor area from the benign area. Segmentation was performed using a combination of feature extraction and protein colocalization algorithms. All epithelia stained with Alexa488-conjugated anti-CK8 and CK18 antibodies, whereas stained basal epithelium Alexa555-conjugated anti-CK5 and Trim29 antibodies ^19,20. Using automatic Definiens (Munich, Germany) image analysis, epithelial structures with an outer layer of basal cells were considered benign, while those without basal cells were considered cancer ²⁰ . Non-epithelial areas were considered stroma. Finally, quantitative biomarker values were extracted from cancer epithelium (“region of interest”; FIGS. 27B-D) only.

In order to evaluate the quality of tissue samples for inclusion in the study, we evaluated the staining intensity of several protein markers in benign tissue. Examination of large numbers of prostate tissue blocks of various qualities indicate that the strength of cytokeratins 8 and 18 and pSTAT3 varies from “high” to “low” or “absent” in benign epithelial regions and vascular endothelium, respectively. (Data not shown; Massimo Roda, personal communication organization). Based on this, we categorized formalin-fixed paraffin embedded (FFPE) prostate cancer tissue blocks into four quality groups (FIG. 27E and Table 8). Only the best two groups of blocks were used to make tumor microarray blocks (TMA), thereby controlling the degradation and differences of biological samples due to preliminary analysis variations ^21-23 . In total, we obtained and tested 508 unique prostatectomy samples with available lethal outcome annotations (Folio Biosciences, Powell, OH). Of these, 418 passed the quality test and were used in our TMA (Table 10).

To balance biomarker signal levels in our multiplex assay format, proteins with high expression levels such as cytokeratin and Trim29 were visualized directly with conjugated antibodies, but with biomarkers with low expression levels Signal amplification through the use of secondary and tertiary antibodies was required. Using test prostate TMA containing low-grade and high-grade tumor material, dilutions of each antibody should be used to minimize background and maximize specificity, as well as low signal values and high Optimized to ensure a dynamic range with at least three times the difference in signal values (FIG. 28B). Signals from serial TMA sections showed higher reproducibility, typical R ² correlation values were above 0.9, and absolute value differences were typically less than 10% (FIG. 28B, data Not shown).

Applicability The present inventors have predictive of lethal outcome the platform was tested using 4 protein index reported in a recent study published by Ding et al ^8. Using a TMA composed of 405 cases obtained with Physician's Health Study (PHS), the authors used a multivariate model based on semiquantitative pathologist-evaluated protein levels of PTEN, SMAD4, CCND1 and SPP1 Showed that a fatal outcome could be predicted. We use our automated platform, not the pathologist, to determine if we can predict a lethal outcome by assessing protein levels in an independent prostatectomy cohort. Asked. Of the 418 eligible cases in our TMA, 340 were found to be useful for analysis, with the exception being mainly due to core misalignment during section preparation (cohort description and PHS (See Figure 28A for comparison with cohort). Quantitative tumor epithelial biomarker levels were extracted from each sample and the values were subjected to univariate analysis. PTEN, SMAD4, and CCND1 were all individually predicted to have a lethal outcome, with hazard ratios (HR) of 2.74, 2.48, and 1.99, respectively, although SPP1 had significant predictive performance. (FIG. 28B).

Next, multivariate Cox and logistic regression analysis was performed. The performance of the 4-marker model was examined as area under the curve (AUC) and concordance index (CI) (FIG. 29A and Table 11 respectively). Logistic regression analysis defined cases as patients who died of prostate cancer. The AUC was approximately 0.75 for the four markers in the training mode and 0.69 to 0.70 in the test mode with logistic regression and Cox analysis, respectively (FIG. 29A). A Kaplan-Meier curve comparing the top one-third of the risk score with the bottom two-thirds based on the four markers was generated by the Cox model trained in all cohorts. This curve shows a clear difference in survival between risk groups (Figure 29B). FIG. 29B shows a comparison of our results with those from testing with PHS. Our average AUC 0.75 [95% confidence interval (0.67, 0.83)] is equal to the average AUC 0.83 [95% confidence interval (0.76, 0.91)] of PHS. It is equivalent to performance. Note the large overlap in the confidence interval.

Incorporating the active state of a protein as part of a multivariate indicator

Since the active state of the protein reflects functional events associated with the aggressive behavior of the tumor, we are not only at the protein level but within post-translational modifications or cellular components by our platform. It was tested whether the activity state of proteins reflected by alterations in localization could be measured quantitatively. Phosphorylation is an example of a well-studied post-translational modification; the stoichiometry of protein phosphorylation at a particular site is an indirect measure of the active state of the parent signaling pathways ^24,25 . Specifically, we can substitute the active state of one or more signaling molecules in the core PTEN-regulated signaling pathways PI3K / Akt and MAPK in place of PTEN in the 4-marker model I examined whether or not. Because the PTEN protein is the only modification in the prostate cancer subset, as opposed to the PI3K / AKT pathway ^11,26 , our objective is to provide a more broad reference for the active state of the PI3K / Akt pathway. ^26,27 which was to identify the replacement phosphomarkers that could be obtained, for this purpose we have made several phosphorylation site specific monoclonal antibodies (P-mAbs) directed against important phosphoproteins. Obtained and tested for technical suitability (Table 7). Testing included specificity analysis by knockdown in cell lines, signal intensity in human prostate cancer tissue, and, importantly, epitope stability ^23,27 based on signal conservation in the entire prostate cancer FFPE sample ( FIG. 30, data not shown). Since the activity of the PI3K / AKT pathway is often independent of PTEN protein states ¹² , ¹³ , we have included phospho-markers. Based on such criteria, the following phosphorylation site specific antibodies were selected and univariate and multivariate lethal outcome prediction performance: p90RSK-T359 / S363, pPRAS40-T246, pS6-S235 / 236 and pGSK3-S21 / 9 (Cell Signaling Technology, Danvers, MA; ²⁷ ). We also selected anti-Foxo3A antibodies for testing. This is because the PI3K pathway is cleared from the nucleus when activated ²⁸ . Markers were subjected to univariate analysis in Kaplan-Meier plot. pPRAS40 and pS6 had significant univariate performance, and the HR was around 2 when comparing the top third signal value to the bottom two thirds (FIG. 28C).

  The inventors then examined the performance of four original markers without PTEN (FIG. 29A). AUC (training) decreased from 0.75 to 0.72 to 0.73. The addition of pS6 (essentially pS6 is substituted for PTEN) increased CI and AUC to 0.75 and 0.76, respectively, but replacement with pPRAS40 significantly increased AUC and CI. Was not provided (data not shown). Finally, replacing PTEN with both pS6 and pPRAS40 increased the AUC (training) value from about 0.76 to about 0.77 (FIG. 29A). The Kaplan-Meier curve corresponding to these three markers combined with pS6 + pPRAS40 is shown in FIG. 29B. These results show that the present inventors have successfully managed the known lethal outcome prediction marker PTEN with the two phospho-markers pS6 and pPRAS40, while maintaining the possibility of predicting the lethal outcome. Indicates that it can be replaced.

Discussion This study has established an automated imaging platform that integrates morphological and proteomic information accurately and reproducibly. We evaluated the performance of the platform by using the same four types of markers reported to predict fatal outcomes by direct comparison with previous studies. A simple meta-analysis of the two trials estimated an insignificant mean AUC difference of 0.08 [95% confidence interval (−0.03, 0.19)]. The difference in performance may be due to methodological differences between the two tests. First, we used monoclonal antibodies whose specificity was verified by siRNA oligo-mediated knockdown in Western blotting and immunohistochemistry (FIG. 30), but 2 out of the antibodies used in the PHS studies. The type is polyclonal and is therefore not ideal for continued anticipatory applications. Furthermore, although the quantitative measurements in this study were fully automated, they could be expected to be somewhat less reproducible overall, as they depended on pathologist interpretation. Finally, in our cohort, predicting lethal outcomes can be more difficult than in high-grade cases, and further predicting lethal outcomes Examples with a high ratio of Gleason ≦ 6, limited by a median follow-up of 11.92 years, not long enough were included. In view of such methodological and AUC differences, our results are comparable, indicating the usefulness of this fully automated platform and prognosis independent of human interpretation.

  In the embodiments featured herein, robust tissue segmentation algorithms and quantitative biomarker measurements are performed in the tumor epithelial region by combining Vectra multispectral image decomposition with a programmable Definiens Tissue Developer. The methods provided herein provide an automated approach that is highly sensitive, can be performed without subjective intervention, and allows very small amounts of cancer tissue to be evaluated successfully.

An example of an important application of the platform of the present invention is the ability to incorporate protein activation states as biomarkers. Here we show that p-mAb, which measures the activity state of signaling molecules in the core PI3K and MAPK pathways, can be substituted for PTEN, an advanced outcome prediction marker. The tumor suppressor PTEN is only modified in 15-20% of early stage prostate cancer, but is often functionally inactivated by various other mechanisms that can be reflected in altered activity of the PI3K / Akt pathway. ¹² Without wishing to be bound by theory, measurement of the active state of the PI3K / AKT pathway may be more informative than PTEN in early prostate cancer lesions. We now show the lethal outcome prediction performance of new 5-marker indicators for radical prostatectomy: SMAD4, CCND1, SPP1, pPRAS40 and pS6.

In summary, we have developed a multiplex proteomics in situ imaging platform with automated objective biomarker measurements that can predict fatal outcomes using prostatectomy tissue, independent of pathologist interpretation. developed. Importantly, the inventors have shown that quantitative measurements of protein activity reflected in post-translational modifications can be incorporated into multivariate protein predictors of lethal outcome. This platform is widely applicable across disease states. In particular, the inventors have already applied this to develop a prognostic prostate cancer biopsy test for early stage lesions, which in many cases have limited tissue mass.

Example 6: Identification and clinical evaluation summary of proteomic biomarkers that predict prostate cancer aggressiveness despite biopsy sampling error This study predicts prostate cancer aggressiveness and lethal outcome despite sampling error Figure 2 shows the identification and clinical evaluation of intact tissue protein biomarkers.

  Determination of prostate cancer aggressiveness and appropriate treatment is based on clinical pathological parameters such as Gleason classification of biopsy material and degree of tumor involvement, prostate specific antigen (PSA) level and patient age. Important difficulties in predicting tumor aggressiveness based on the Gleason classification of biopsy materials include prostate cancer heterogeneity, biopsy sampling error and biopsy interpretation variability. The resulting risk assessment uncertainty results in significant overtreatment, accompanied by costs and unhealthy rates. We have developed a performance-based strategy to identify protein biomarkers that are tailored to more accurately reflect true prostate cancer aggressiveness despite differences in biopsy sampling. Prostatectomy samples with pathological and lethal outcome annotations from a long-term follow-up large patient cohort were blindly evaluated by a specialist pathologist who identified the tissue region with the highest and lowest Gleason classification for each patient . To simulate biopsy sampling errors, "high" and "low" tumor microarrays were made using cores from the high and low Gleason regions of each patient sample, respectively. Using a quantitative in situ proteomics approach, we have determined that from among the 160 candidates, the most prominent prostate cancer aggressiveness (surgical Gleason score and pathological TNM progression classification) and lethal outcomes in the high Gleason region and Twelve biomarkers were identified that predicted robustly in both low Gleason regions. Conversely, the previously reported lethal outcome marker markers for prostatectomy tissue did not work in the context of maximum sampling error. Our study has led to the discovery of cancer biomarkers that are resistant to sampling errors in biopsy material where prostate cancer pathology is predicted during biopsy for general and clinical trials.

INTRODUCTION Prostate cancer accounts for 27% of cancer cases diagnosed in men in the United States, and according to estimates by the American Cancer Society, 233,000 prostate cancer cases were newly diagnosed in Japan in 2014 (1 ). Although the risk of death from prostate cancer has been significantly reduced as a result of early detection and improved treatment options (1), there are overdiagnosis and overtreatment concerns for this common cancer (2,3) . Of the newly diagnosed cases, about 1 in 7 cases of prostate cancer progress to metastatic disease during their lifetime, but about half of newly diagnosed men with prostate cancer progress. People with localized disease who have a very low risk of (1,4). Despite this low risk, as many as 90% of men diagnosed with low-risk prostate cancer in the United States receive radical treatment, usually radical prostatectomy or ablation radiation therapy (5). In diseases that are difficult to reveal clinically, such treatment can be excessive, often resulting in long-term adverse events such as urinary incontinence and erectile and bowel dysfunction (2, 6, 7) .

  Current guidelines and recognized standard care for the diagnosis and management of prostate cancer recommend using clinical and pathological parameters to assess disease grade and progression classification at biopsy (8, 9). Pathological evaluation of the tissue obtained by needle biopsy is essential for both confirmation of prostate cancer diagnosis and cancer malignancy classification. The grade determined by the biopsy Gleason score (GS) is the most important predictor of outcome and is most helpful as a guide for management judgment. Approximately 80-85% of all prostate cancer biopsies have a GS 3 + 3 = 6 or 3 + 4 = 7 representing a series of cases at risk of progression from low to moderate to high (10). Patients suspected of having indolent disease are candidates for active surveillance (3, 8, 9). However, current biopsy assessment methods often cannot accurately sort individual patients according to this range (5, 10).

  There are two factors recognized to affect the accuracy of biopsy-based Gleason scoring: one is the sampling difference (ie, the region with the highest Gleason classification cannot be sampled), the second Is inconsistent pathologist Gleason scoring (10-12). Despite the current standard practice of multi-core biopsy sampling, the most aggressive areas of the tumor are frequently underpresented or overpresented (11, 13). Indeed, 25-50% of prostate cancer cases need to be graded from initial biopsy score to more accurate surgical GS or reduced in grade after analysis and classification of prostatectomy tissue (10, 14, 15). The discrepancy between pathologists in the Gleason classification stems from the subjective aspect of the Gleason scoring system, which is particularly applied to small samples. Such discrepancies make it difficult to ensure a uniform and accurate prognosis and can be as high as 30% (16, 17).

  Several clinical and pathological risk stratification systems, such as the D'Amico classification system, Cancer of the Prostate Risk Assessment (CAPRA) score, and the National Comprehensive Cancer Center Network to improve the prediction of prostate cancer aggressiveness (NCCN) guidelines have been developed. (9, 18-20). All such systems recognize that biopsy GS is the single most powerful and beneficial in risk assessment. The GS is composed of two Gleason patterns, with the dominant pattern first manifesting. The two are added together to determine the Gleason score. According to the 2005 consensus on Gleason scoring, only three patterns (3, 4, and 5) are typically found in biopsies (21). The acceptable prognostic category GS is 3 + 3 = 6, 3 + 4 = 7, 4 + 3 = 7, 8, and 9-10. Importantly, 3 + 4 = 7 and 4 + 3 = 7 have comparable Gleason sums, but the latter has a significantly worse prognosis based on the higher amount of pattern 4 (16, 22). Importantly, all risk stratification systems used to guide clinical management rely on effective and consistent Gleason scoring and are therefore prone to sampling differences and pathologist scoring discrepancies .

  Enhancement of biopsy strategies has been proposed as one means of resolving sampling differences and errors. Among these, increasing the number or density of core samples may ensure more representative collection of tumor tissue. However, increasing the number of biopsy samples collected to more than the 12 currently recommended may increase the risk of adverse events due to oversampling and evidence that the pathological classification is improved by increasing it There are few (23, 24). There is also interest in a new form of image-guided biopsy. Currently, MRI-guided biopsy appears to improve the detection of aggressive cancer, but long-term trials are needed to determine whether MRI can improve patient choice for active surveillance (AS) (25)

  Using a quantitative multiplex proteomics in situ imaging system (26) that allows accurate biomarker measurements from intact tumor epithelium, we have now demonstrated prostate cancer aggressiveness (prostatectomy (surgical)). Report the identification and evaluation of twelve biomarkers that can predict fatal outcomes as defined by Gleason score and pathological TNM progression classification. The marker was specifically selected to be robust against sampling error. The study was performed on prostatectomy tissue and included a simulation of sampling bias error in biopsy material based on core collection from each patient's high and low GS regions. This approach could be used to reflect true prostate pathology as determined by prostatectomy GS and pathological progression classification, whether measured in the high or low score Gleason area Biomarkers were selected based on In addition to reflecting aggressive disease states, biomarker candidates were also evaluated for their potential to predict prostate cancer-specific mortality in the low- and high-grade areas of heterogeneous cancer. This performance-based approach identified novel biomarkers and identified known biomarkers that predicted prostate cancer aggressiveness and lethal outcome.

Results Biopsy Simulation A biopsy sampling model was developed to simulate and exaggerate the diversity of biopsy samples observed in clinical practice. For this purpose, we embedded cores from annotated prostatectomy tissue in a tissue microarray (TMA). Based on intensive Gleason classification by a specialist urologist pathologist, the core for each patient is taken from the area with the lowest aggressive tumor (low GS) and embedded in a low grade tissue microarray (LTMA); In parallel, cores were taken from the areas with the most aggressive tumors based on Gleason classification (high GS) and embedded in high-grade tissue microarrays (HTMA) (FIG. 32). Accordingly, the inventors have developed a paired tissue TMA with a bi-biased sample that represents both high and low invasive tumor areas for each patient.

  Table 12a shows the clinical signs of a multicenter cohort of 380 patients who made paired TMA. Table 12b shows the corresponding core Gleason and its surgical (prostatectomy) Gleason in HTMA, along with a subset of 301 cases with core Gleason 3 + 3 or 3 + 4 in LTMA.

Tables 12a and 12b show the clinical signs of the cohorts used to make the L and HTMA.

  Sampling for LTMA was specifically designed to underestimate the severity of the disease. As shown in Table 12a and Table 12b, 64.7% of the L TMA samples had a core GS of 6 or lower, but of these L TMA samples, the surgical GS of 6 or lower. Only 30% were from patients with For samples of LTMA from cases with high surgical GS from core GS ≦ 3 + 4, the probability of increased malignancy (Table 12b) is 0.64 (95% Wilson confidence interval [CI]: 0.59-0.69). )Met. This probability of increasing malignancy is higher than that seen in clinical practice (12), as expected from the sampling method used and the patient cohort. Therefore, by exaggerating the sample diversity expected in clinical practice, this biopsy simulation procedure will identify biomarkers that are highly reliable in predicting prostate cancer invasion regardless of sample diversity A useful model was obtained.

Effect of sampling error on known biomarker model performance To evaluate the effect of sampling differences on prognostic marker performance, we first diagnosed a fatal outcome when used on prostatectomy tissue. The established biomarker combinations reported to be tested for the possibility of predicting lethal outcomes and aggressive disease when used against biopsy simulated tissue. Previous studies have shown that radical prostatectomy (RP) GS of 7 or higher and prostate cancer expansion beyond the glandular portion of the prostate are significant predictors of metastasis and prostate cancer-specific mortality ( 27-29). Therefore, we defined an “aggressive disease” as surgical GS at least 3 + 4 or pT3b (seminal vesicle invasion), N + or M + based on prostate pathology. We have reported the four biomarker models reported by Ding et al. (30) (SMAD4, CCND1, SPP1, PTEN) in disease-specific death and disease invasion in a TMA cohort with our sampling differences. We tested the possibility of predicting both sexes. The patient core in L or H TMA was divided into separate “training” and “test” datasets, and logistic regression models were used to estimate marker coefficients using the training dataset. We estimated the area under the curve (AUC) from the receiver operating characteristics (ROC) obtained in the test set and then repeated this process with further sampling. As shown in Table 13, when measured with HTMA, disease-specific mortality could be predicted with this 4-marker index, with a median test AUC of 0.65 (95% CI 0.59-0 .74). However, as measured by LTMA, which represents an underestimation of surgical GS, this 4-marker model showed an insignificant test AUC median of 0.49 (95% CI was 0.42-0.58). Furthermore, the possibility of predicting aggressive disease with this 4-marker index when measured with either H or L TMA was not significantly reached (median test AUC was 0.56 [95% CI was 0.44-0.64] and 0.56 [95% CI is 0.46-0.65]). Such results indicate the impact of sampling error on prognostic marker performance and the importance of identifying alternative biomarker combinations that can accurately predict outcomes despite such sampling differences.

Biomarker identification

  After showing that the bias biopsy simulation TMA actually reflects an extreme sampling error scenario and that the known predictive marker index cannot be reproducibly performed due to such sampling differences, the present inventor then And others sought the primary goal of identifying biomarkers that could be robustly predicted to be aggressive regardless of biopsy sampling differences. By utilizing prostatectomy tissue samples with abundant clinical and pathological annotations from a large cohort of long-term follow-up patients, we have established a performance-based strategy for selecting potential markers did. The stepwise approach included: 1) identification of the candidate biomarker, 2) assessment of its biological and technical suitability, and 3) analysis of performance in the H and L TMA cohorts (Figure 33).

  The process began with a search of gene expression datasets available to published literature and the public, which identified 160 biomarker candidates based on biological relevance for prostate cancer (30-48). We further determined 120 of these priorities based on the availability of appropriate monoclonal antibodies (MAbs) (see Table 14 for a comprehensive list of biomarker candidates). Candidates include well-characterized markers associated with prostate cancer aggressiveness, such as EZH2, MTDH, FOXA1 (49-51), and markers PTEN, SMAD4, Cyclin D1, SPP1, phospho-PRAS40-T246 ( pPRAS40) and phospho-S6-Ser235 / 236 (pS6) (previously identified as predicting a lethal outcome in prostatectomy tissue) (26, 30) were included.

  Next, we obtained Mabs for 120 prioritized candidate biomarkers and tested their specificity and suitability for a quantitative multiplex immunofluorescence (QMIF) assay. Candidate MAbs for further analysis were selected based on signal intensity and specific immunofluorescence (IF) staining patterns as described elsewhere (26). We determined the priority of MAbs that preferentially stained cancer cells over stromal cells. Based on the large number of stained samples, we observed that the IF staining intensity of the epithelial marker is low in tissues that appear to be poorly fixed or preserved. Candidate biomarker antibodies were selected based on signals that were more stable than those of epithelial markers.

  In the third step, we tested 62 MAbs that passed the previous step and examined their dynamic range and predicted performance. Using a small test TMA designed to represent the lowest aggressive area of prostate tumors with high overall GS and low overall GS, biomarkers are selected based on the correlation between signal intensity and surgical GS did. Specifically, the inventors have shown a difference in the distribution of signal values between non-aggressive cases and aggressive cases, as well as a triple signal between the lowest expression value and the highest expression value. Needed a difference. The final 39 candidate MAbs meeting this criteria were tested in the clinical cohort represented by the H and L TMA blocks described above.

Univariate analysis Our next objective was to further evaluate candidate biomarkers based on univariate prognostic capabilities and analysis performance in the context of sampling error. Each of the 39 biomarkers identified above is disease aggressiveness (surgical GS ≧ 3 + 4 or pathological progression classification pT3b and / or N + as measured in both low and high Gleason regions Or M +) and the possibility of predicting death (survival analysis) due to the disease. Individual markers labeled with two asterisks are predictive (P) for aggressive disease or death from prostate cancer based on an increase or decrease in expression regardless of whether measured in the low or high Gleason region. <0.1) was shown (FIG. 34). This result suggests that these markers are resistant to varying degrees of sampling error. If you measure only in the high Gleason area and not in the low Gleason area, there are 2 markers for predicting aggressiveness and 3 markers for lethal outcome, and these markers are not robust to sampling errors It has been shown. Conversely, the inventors were unable to identify markers with predictive performance when measured in the low Gleason region but not in the high Gleason region. Notably, the strong association between aggressive disease and lethal outcome is that 12 of 14 markers with significant univariate performance for aggressiveness are significant for lethal outcome. The finding of univariate performance also made it clear. As a further validation of the performance-based biomarker selection approach, we found a strong correlation between the lethal outcome and the expression of three known prostate cancer progression markers EZH2, HoxB13 and MTDH2 (49-51) It was confirmed. In conclusion, we have identified several candidate markers that have univariate performance for both aggressive disease and lethal outcome and are also robust to sampling errors. In addition, the inventors have identified markers that are predictive only in situations where sampling error is minimal (performance with HTMA but not LTMA).

Multivariate Analysis: Biomarkers that Predict Tumor Invasion In order to explore the best multivariate biomarker combinations that are predictive of disease aggressiveness, we have identified all five possible models. Using the biomarker combinations up to (including), a thorough search was performed (FIG. 35A). In the multivariate analysis, attention was paid to 31 types of biomarkers selected based on technical criteria such as MAb detection signal intensity, dynamic range and specificity from the original 39 types of sets (see Materials and Methods). Initially, an “extreme” model approach was used for multivariate analysis, including excluding intermediate samples (GS = 3 + 4, ≦ T3a and N0) in model building and testing. We split the patient core in L TMA into independent training and test case sets and test the resulting model for multivariate performance under sampling differences in both L TMA and H TMA. did. For this purpose, we use a logistic regression model to estimate biomarker coefficients using a training data set, estimate AUC from the ROC obtained in the test set, and then separate this process from another Repeated with sampling.

In each case, the biomarkers most frequently present in the top 5% or 1% in the model, sorted by AIC (Akaike Information Criterion) (52) and test set AUC were examined. Final tally was created for ranking by test, ranking by AIC and both rankings (see FIG. 35B for a representative example of 5 biomarker models ranked by AIC and testing). We observed a high degree of preservation of biomarker ranking in the top performance biomarker model (see FIG. 35C and Table 15). Among them, the following biomarkers: ACTN1, FUS, SMAD2, DERL1, YBX1, DEC1, pS6, HSPA9, HOXB13, PDSS2, SMAD4, CD75 appeared to be top markers in at least 50% of the ranked list. CUL2 was also present in several high ranking models (see Table 15 for further details of ranking results).

Multivariate analysis: biomarkers predicting lethal outcomes A similar model design analysis was performed for lethal outcomes (Table 16). Among them, biomarkers considered to be top markers in at least 50% of the ranked list: MTDH2, ACTN1, COX6C, YBX1, SMAD2, DERL1, CD75, FUS, LMO7, PDSS2, FAK1, SMAD4, DEC1 (ranking results) (See Table 16 for further details).

Final biomarker set

  Based on careful integration of univariate and multivariate performance and analysis considerations such as a minimum three-fold signal intensity dynamic range in tumor samples for all antibodies, the final 12 biomarkers Select a pair. FIG. 36A shows the estimated odds ratios (OR) associated with these 12 biomarkers for aggressive univariate predictions, and for each biologic based on both univariate and multivariate analyses. The basis for selecting markers is summarized. FIG. 36B shows a biological overview of the selected biomarkers. The final biomarker set consisted of: FUS, PDSS2, DERL1, HSPA9, PLAG1, SMAD2, VDAC1, CUL2, YXB1, pS6, SMAD4, ACTN1.

  Each of these 12 marker antibodies was strictly verified by specificity analysis including Western blotting (WB) and immunohistochemistry (IHC) assays, as shown in FIG. 37, before and after target-specific knockdown. did. Interestingly, by this process, we found that MAb sold as specific for DCC detects HSPA9 (also known as mortalin) but not DCC (FIG. 38). Since DCC knockdown did not result in the disappearance of the only band on WB, we performed mass spectrometry sequencing analysis and identified this protein as HSPA9. The fact that this protein has a well-reported role in cancer progression and survival (53) further validates performance-based and function-based biomarker identification approaches.

  Next, we used the model design approach described above to determine the potential for predicting the final 12 biomarker set for both disease aggressiveness and disease-specific mortality in any patient cohort. evaluated. RTMA and HTMA data are randomly distributed into training and test case sets, logistic regression is performed on the LTMA training case sets, performance is evaluated on the LTMA and HTMA test case sets, Repeatedly, a 12 marker model for disease invasion was developed. As shown in FIG. 36C, this resulted in an L TMA test AUC of 0.72 (95% CI: 0.64 to 0.79) and a corresponding aggressive disease OR of 20 / unit of change in risk score (95% CI: 4.3-257). To confirm the possibility of generalization under sampling error, models derived from the L TMA training case set were also tested for prediction of aggressive disease in HTMA and the results were consistent (FIG. 36C). Without making any further changes to this aggressive model, we examined its performance for predicting lethal outcome by correlating the aggressive risk score with death from disease. Notably, we found that the AUC of lethal outcome was similar to that of aggressive in both LTMA and HTMA (respectively 0.72 (95% CI: 0 .60-0.83) and 0.71 (95% CI: 0.61-0.81)). The corresponding HR for lethal outcome in LTMA and HTMA was 66 / variation units (95% CI: 5.1-6756) and 36 (95% CI: 3.3-2889), respectively. We conclude that the twelve identified biomarkers are robust to sampling error and likely predict both aggressive and lethal outcome of the disease.

Discussion Prostate cancer aggressiveness at the time of initial diagnosis and as part of ongoing follow-up of patients, for example, patients assigned to active patient monitoring and patients receiving active treatment for this disease As such, there is an ongoing clinical need for more accurate assessment (4, 29, 54). Currently, in men with early disease, a biopsy GS value of 3 + 4 = 7 or greater is one of the useful prognostic factors that indicate the need for aggressive treatment (9,55). As such, biopsy sampling errors due to tumor heterogeneity and Gleason scoring discrepancies can affect the accuracy and reliability of risk assessment for cancer progression, aggressive and lethal patients. This uncertainty contributes to the situation that prostate cancer is significantly overtreated because it is difficult to accurately predict the prognosis of patients with biopsy material Gleason classification of 3 + 3 or 3 + 4. (2, 5, 10, 54, 56, 57).

Biomarkers for predicting aggressiveness and lethality of prostate cancer In this specification, prostate cancer can be detected even in the context of extreme sampling differences, a problem typically seen when collecting prostate biopsies. Describe the successful development of a performance-based method for identifying and evaluating biomarkers with predictive aggressive and lethal outcomes. Using a large cohort of annotated clinical prostatectomy samples with long-term follow-up for lethal outcome (N = 380), a specialized pathologist in a blinded manner in the area of highest and lowest GS on each prostatectomy tissue Marked. With such a core collection of “high” and “low” regions from each patient sample, we create a paired TMA that represents the entire cohort, thereby creating a biopsy with sampling error for each patient. The material was simulated. Using such a paired TMA, we have determined that aggressive prostates when large numbers of biomarker candidates are measured in either the low-grade or high-grade cancer area of each patient. We assessed the possibility of predicting the pathophysiology and fatal outcomes. Specifically, we first select biomarkers that have performance against aggressive and lethal outcomes when measured in LTMA tissue and are most robust against extreme sampling errors Was identified. For this purpose, we included only LTMA samples with core Gleason ≦ 3 + 4 as clinically relevant. This is because biopsies with GS of 4 + 3 or higher are inevitably candidates for active treatment. Biomarker candidates were quantified using an integrated multiplex proteomics in situ imaging platform that provided automated objective biomarker measurements (26). Based on univariate analysis, most of the identified biomarkers predict both disease aggressiveness and prostate cancer-specific mortality, whether measured with LTMA or HTMA tissue samples And therefore robust against sampling differences (FIG. 34). In addition, several prostate cancer biomarkers previously reported to predict progression risk and lethal outcome, such as SMAD4, EZH2, MTDH2, HoxB13 and PTEN, were all confirmed positive for lethal outcome, Support the verification of this approach.

  As part of our antibody specificity verification, we have found that MAb sold as anti-DCC is actually irrelevant by targeted knockdown analysis and mass spectrometry protein sequencing analysis. It was found that the protein recognizes either HSPA9 or mortalin. We have found that HSPA9 is a predictor as part of the multivariate model and is thus included in the final 12 marker sets. When subjected to functional analysis, the inventors found that HSPA9 was actually involved in clonogenic cell colony assay configuration and cell proliferation, consistent with previous findings (FIGS. 38 and (53)). reference). This further supports verification of an unbiased performance-based marker selection approach.

  Twelve biomarkers were selected based on univariate performance and marker frequency in a multivariate model of disease aggressiveness and lethal outcome (FIG. 36A). Multivariate models based on these 12 markers showed similar predictive performance for aggressiveness under differences in tissue sampling (FIG. 36C). Interestingly, risk scores generated based on this 12-marker aggressiveness model equally predicted separate endpoints of lethal outcome under differences in tissue sampling (FIG. 36C). The fact that markers predicting prostate cancer aggressiveness, as defined by prostate Gleason classification and progression classification, also predict fatal outcomes, strongly links the aggressive characteristics and lethality in surgical pathology Supported. More importantly, the use of our pathological endpoint as relevant to long-term patient outcomes for the construction of our biomarker panel is validated. The twelve identified biomarkers are associated with predicting tumor behavior and provide the basis for clinical evidence-based multivariate biopsy testing to assess prostate cancer aggressiveness. Biopsy plays an important role at initial diagnosis and in monitoring the disease state of patients undergoing active monitoring (8, 9). As such, the multivariate biopsy test described herein can provide information on early decision processes in the management of patients with prostate cancer.

Biomarkers that are robust to sampling errors The tests of the present invention have identified and selected markers that are highly robust to sampling errors. One of the main reasons for biopsy sampling errors is the heterogeneity of prostate cancer. The inability to consistently acquire tissue from the most aggressive part of the tumor frequently results in an underestimation of tumor aggressiveness and progression risk. By coring at the highest and lowest Gleason regions of each patient, we have two biopsies for each patient (one with the “maximum” sampling error (L TMA), the other with the minimum) A paired TMA was created for all cohort tests designed to simulate those with a sampling error of (H TMA). The inventors have focused on LTMA with core Gleason ≦ 3 + 4. This is because it represents a clinical case where standard medical care is inadequate for accurate prognosis. We find that about 54% of such LTMA cases are graded up to higher surgical Gleason scores (which are higher than those observed in clinical practice) (12 ) Confirmed that our approach provides a bias sampling error model (Table 12b).

  The need for identification of biomarkers that are tolerant of sampling errors is a well-established 4 based on cyclins D1, SMAD4, PTEN and SPP1, previously reported to predict fatal outcome based on prostatectomy cohort Emphasized by examining the marker index (30). While this model showed predictability of lethal outcomes with HTMA representing the situation of minimal sampling error, this model is quite lethal with our LTMA tissue core representing the greatest sampling error. (Table 13). This finding is consistent with recent reports that this 4-marker index cannot predict fatal outcome in low Gleason score prostate tumors (58).

Based on univariate marker analysis, we identified 14 and 18 markers with robust performance in sampling errors for disease aggressive and lethal outcomes in LTMA and HTMA samples, respectively ( (Marker with ^** displayed in FIG. 34). Most of these univariate selectable markers predict both indications under sampling differences, again correlating between disease aggressiveness and lethal outcome as discussed above for the above multivariate analysis Is supported. Interestingly, all markers that showed univariate performance for disease aggressiveness and lethal outcome in LTMA were also predictive in HTMA, but two markers (PXN and MTDH2) and three The markers (NCOA2, CCND1 (cyclin D1) and AKAP8), respectively, showed predictability of aggressive disease and lethal outcome only when measured with HTMA, and not with LTMA (FIG. 34). This indicates that these markers are predictive primarily in situations where sampling error is minimal. In fact, all five of these markers have been shown to be important regulators of cell proliferation, migration and carcinogenesis (see, eg, (30, 51, 59-61)). The observation that cyclin D1 is predicted to have a lethal outcome in HTMA alone and not in LTMA is the result of the 4 marker index reported by Ding et al. In our LTMA tissue and low grade prostate cancer samples. Consistent with the finding that it does not show predictability of lethal outcomes (58). The fact that there is no marker that predicts aggressive disease or lethal outcome in LTMA alone and not in HTMA, we primarily predict either aggressive or lethal outcome in LTMA It is interesting to consider that the marker that reflects the maximum sampling error robustness can be selected. This indicates that the identified marker probably reflects the field effect of more aggressive tumor areas, consistent with its similar performance in LTMA and HTMA tissues.

Genetic and proteomic approaches In search to find better new biomarkers in prostate cancer, there is great importance and progress in identifying potential genetic markers that can provide information on clinical risk prognosis (31, 32, 39, 48, 62, 63). However, many of the genes identified have inconsistencies or inadequacies in the reliability results of such markers in disease prognosis. For example, the TMPRSS2-ERG fusion gene has been reported to be associated with high-risk tumors, but in a very recent study using a large cohort, there is no strong correlation between this fusion type and patient outcome (64). In a multivariate gene expression system study, recently, metastatic disease and fatal outcomes have been observed following biochemical recurrence after treatment in a conservatively managed patient cohort in the UK (65) and in an aggressively managed cohort in the US (66,67). It is reported that it is predicted based on The impact of sampling differences on this study has not yet been established.

  The results of this study show that the use of a proteomic approach that measures proteins only from the tumor area of intact tissue can improve the accurate risk classification in the progression classification at biopsy. The rationale for this idea is two-factor. First, because prostate cancer is a heterogeneous multifocal disease, the biopsy material frequently contains only low-grade components and can be classified as low-risk cancer by a pathologist. However, it has been reported that high-grade molecular features that are not reflected morphologically have spread throughout the cancer (68, 69), and therefore can be measured even in biopsy materials that appear to contain low-grade grades. It has become. Proteomic approach to measure proteins from intact tumor areas only, sensitive to these high-grade molecular features in situ, even in tissue samples with varying amounts of tumor components relative to benign components It is possible to evaluate with. This is an advantage over gene expression technology that requires tissue homogenization and dilutes the high-grade molecular features differently depending on the amount of benign tissue present. Secondly, Gleason classification for biopsy is subjective and there is no agreement among specialist pathologists in up to 30% of cases (16, 17). Risk classification is improved by molecular attributes that can be measured objectively.

  The twelve biomarkers identified in this study are proteins with a range of functions, including transcription, protein synthesis, and regulation of cell proliferation and apoptosis and cell structure (30). The fact that the biomarker can function despite biopsy sampling errors means that protein biomarkers can further improve Gleason risk classification as a means to guide early management of prostate cancer treatment Indicates.

Conclusions Given the difficulty in predicting survival outcomes for patients diagnosed with early-stage cancer and the resulting overtreatment, there is an urgent need for reliable and accurate prognostic testing for patients with prostate cancer. ing. The protein biomarker identification strategies described herein can also be applied to other tumor types, and other tumor prognostic or predictive tests where histological assessment is critical for risk stratification and prognosis A performance-based selection of biomarkers that can be used to develop is possible.

Materials and Methods Reagents and Antibodies All antibodies and reagents used in this study were obtained from commercial sources listed in Table 17. Anti-fluorescein isothiocyanate (FITC) MAb-Alexa 568, anti-CK8-Alexa 488, anti-CK18-Alexa 488, anti-CK5-Alexa 555 and anti-Trim29-Alexa 555 are combined with an Alexa dye and an appropriate protein conjugation. Conjugation was performed using a kit (Life Technologies).

Slide Processing and Staining Protocol From the TMA block, 5 μm sections were cut and placed on Histogrip (Life Technologies) -coated slides and processed as previously described (material supplement). Briefly, after deparaffinization, antigen activation was performed in a 0.05% cytocolanic acid solution at 95 ° C. for 45 minutes using a Lab Vision PT module (Thermo Scientific). Staining was done either manually or in an automated manner using Autostainer 360 or 720 (Thermo Scientific).

A QMIF staining procedure using two anti-biomarker antibodies and a region of interest marker in combination was performed as previously described (see Materials Supplements and Methods). For diaminobenzidine (DAB) -based IHC staining, slides with tissue were processed as described above, blocked with Snipereagent ^™ (Biocare Medical), and incubated with the primary antibody solution. UltraVision (Thermo Scientific) was used as a secondary reagent. Finally, the tissue was counterstained with hematoxylin and a cover glass was placed.

Acquisition, processing, quality control and annotation of FFPE prostate cancer tissue blocks A set of FFPE human prostate cancer tissue blocks with clinical annotations and long-term patient outcome information was obtained from Folio Biosciences. Samples were collected with appropriate institutional ethics committee approval and all patient records were anonymized. For evaluation of candidate biomarker antibodies, FFPE human prostate cancer tissue blocks with limited clinical annotation were obtained from other commercial sources.

  A series of 5 μm sections were cut from each FFPE block. For annotation, the last 5 μm sections cut from each FFPE block were stained with hematoxylin and eosin (H & E) and scanned using the ScanScope XT system (Aperio). Scanned images were viewed remotely, and GS was annotated in a blinded fashion by a specialized qualified clinical anatomical pathologist. The four regions with the highest Gleason score pattern and the two regions with the lowest values were marked with annotated circles corresponding to a 1 mm diameter core (FIG. 32, see above).

Preparation of TMA block A TMA block was prepared using the modified agarose block procedure (70). To create the test TMA (MPTMA10), we selected 72 FFPE tissue blocks of prostatectomy samples with annotations available for GS and pathological progression classification, of which 37 Examples had GS 3 + 3 = 6 and T2 progress classification, while 35 had GS 4 + 3 = 7 or 3 + 3 = 6 or 3 + 4 = 7 GS and T3b progress classification. One 1 mm core per patient sample was taken from the lowest Gleason pattern area and placed in the acceptor block.

For the construction of HTMA and LTMA, we used a cohort of FFPE human prostate cancer tissue blocks with clinical annotations and long-term patient outcome information. For each patient sample, the core was taken from the area with the highest Gleason pattern and stored in the H acceptor block. The second core was then taken from the region with the lowest Gleason pattern and placed in the L acceptor block. The arrangement order of the sample cores in the H block was randomly selected, and the positions of the cores in the L block were the same as those in the H block. Also, the cell line control (Table 18) FFPE block cores were placed in the upper and lower portions of all HTMA blocks and LTMA blocks. When finished, 5 μm serial sections were cut from each block and representative sections were stained with H & E and scanned with the ScanScope XT system. The image of the H & E stained core was then independently annotated in a blinded fashion by a qualified anatomical pathologist about the observed Gleason pattern.

  The resulting H TMA and L TMA blocks were identical for a set of patient samples, but the Gleason pattern that could be observed was different (Figure 32, bottom). For this study, two pairs of TMA blocks (MPTMAF5H and 5L, 6H and 6L) were made using cores from 380 patient samples.

Biomarker Selection In order to identify biomarkers for prostate cancer aggressiveness, the inventors have developed selection / evaluation methods that may be widely applicable to diseases and medical conditions. This method shown in FIG. 33 has a biological stage, a technical stage, a performance stage, and a verification stage.

  In the biological stage, an initial list of potential biomarkers for prostate cancer aggressiveness was compiled from publicly available data. The priority of this list was then determined based on biological relevance, in silico analysis, Human Protein Atlas review (www.proteinatlas.org) and the required MAb commercial availability. Investigation of biological relevance was based on the mechanism of action within cells, particularly within disease cells. In silico analysis was based on previously known gene amplifications, deletions and mutations, and the correlation between univariate performance or progression between such genetic modifications and disease. Human Protein Atlas contains data on protein expression levels in various tissues in disease states.

  In the technical stage, commercial MAbs were obtained and tested for the possibility of detecting biomarkers from clinical samples. Initially, we stained samples of malignant and benign prostate tissue using a DAB-based IHC staining procedure, and selected candidate antibodies specific for epithelial cell staining that showed a good signal-to-noise ratio. The inventors have further shown that successful candidates for malignant and benign prostate tissue samples are epithelial cytokeratins CK8 and CK18 and basal markers CK5 and Trim29, which are markers of interest described with IF (material supplements and methods). Were used to test. Antibodies and biomarkers that meet IF criteria were advanced to the performance stage.

  In the performance stage, MAbs were tested in TMA. Performance was evaluated for univariate correlation between tumor epithelial expression and disease state. MAbs and biomarkers showing univariate correlation between expression and disease state were then evaluated for both univariate correlation and performance in combination with other markers in large HTMA and LTMA sets.

Image Acquisition Two Vectra Intelligent Slide Analysis Systems (PerkinElmer) were used for automatic image acquisition as described (material and method supplements). Multispectral images were processed into separate images for each fluorophore signal and sent with a Definiens Developer script (Definiens AG) for analysis.

Defienens Automated Image Analysis We have developed an automated image analysis algorithm using Defienens Developer XD for tumor identification and biomarker quantification. Two 20 × image fields were acquired for each 1.0 mm TMA core. The Vectra multispectral image file was first converted to a multilayer TIFF file using inForm (PerkinElmer), and then the customized spectral library was converted to a single layer TIFF file using BioFormats (OME). Import a single-layer TIFF file into the Definiens workspace using a customized import algorithm so that for each TMA core, both image view TIFF files are loaded as "maps" into a single "scene" and analyzed I made it.

  Automatic adaptive binarization was used to define a fluorescence intensity cutoff for tissue segmentation with each individual tissue sample in our image analysis algorithm. Cell line control cores within TMA were automatically identified based on predetermined core coordinates in the Definiens algorithm. Tissue samples were segmented for classification into epithelial cells, basal cells and stroma using fluorescent epithelial cell markers and basal cell markers with 4 ', 6-diamidino-2-phenylindole (DAPI), and further cytoplasmic And compartmentalized into nuclei. Individual gland areas were classified as malignant or benign in combination with object-related features, such as gland thickness, based on relevant features between basal cells and adjacent epithelial structures. There was no epithelial marker in all cell lines and therefore cell line controls were segmented using autofluorescent channels in tissue versus background. Fields with artifact staining, insufficient epithelial tissue or defocused images were removed by a rigorous multi-parameter quality control algorithm.

  The intensity of epithelial markers and DAPI was quantified as a quality control measure in malignant and non-malignant epithelial areas. The intensity level of the biomarker was measured based on a predetermined subcellular localization criterion in the cytoplasm, nucleus or whole cell of the malignant tissue. The average biomarker pixel intensity in the malignant compartment was averaged in a map with acceptable quality parameters to obtain a single value for each tissue sample and cell line control core.

Endpoints in data stratification and analysis The expression of 39 biomarkers was examined for correlation with tumor aggressiveness and lethality using HTMA and LTMA. Disease aggressiveness was defined based on prostate pathology (aggressive disease = surgical Gleason ≧ 3 + 4 or T3b, N +, or M +). In the aggressive analysis, we examined marker correlation based on measurements on both L TMA samples with core Gleason ≦ 3 + 4 and corresponding matched H TMA samples.

  In the analysis of lethal outcome, we created two different sample sets: (1) all cores with measurements GS ≦ 3 + 4; and (2) all cores.

Cohort composition Table 12a shows the cohort composition. Only samples with a complete set of clinical information were included. When analyzing using a specific set of biomarkers, only samples with values for the markers were considered. Therefore, the numerical value in the table is the upper limit.

Invasive and lethal univariate analysis Our goal of univariate analysis is to characterize univariate behavior as a performance evaluation for potential inclusion in the final marker set, and an exhaustive multivariate model It is a two-factor of obtaining a small set of markers for the search of. All model designs were done in R 3.0 and used standard functions and packages such as glm, survival, KMsurv, binom and pROC. Biomarkers were evaluated based on two outcomes: surgical GS prediction and death prediction (lethality). Inferior or more severely categorized surgical GS predictions were modeled using both OR (logistic regression) and biomarker tools (linear regression). The lethality was modeled using HR (conventional Cox proportional hazards), OR (logistic regression) and marker means (linear regression). Also, Wilcoxon and permutation tests were applied to obtain a nonparametric and robust evaluation.

  34A-B show the key results. In addition, the univariate results were directly considered in the selection of the final marker set as shown in FIG. 36A.

Biomarker ranking for aggressiveness by exhaustive search of multi-marker models We ranked biomarkers by importance in multi-marker models; 31 were selected from the original 39 sets to further improve technical performance 31 Types of biomarkers were also used for exhaustive biomarker searches. We have a bootstrap method for each biomarker combination that takes into account all combinations of up to 5 biomarkers from 31 biomarkers tested with L TMA in HTMA and LTMA analysis. Created a set of 500 training case examples and obtained a related complement test case set. A logistic regression model was applied to each training case set and then tested with each related test case set. Training and test AUCs (ie, statistics) and training AIC were obtained each time. Median and 95% CI were obtained for all three statistics.

  We then considered the biomarker selection frequency in the model and sorted by its AIC and separately by its test AUC. For each ranking obtained in the model, biomarker usage frequencies within the top 1% and top 5% of the list were examined. Biomarkers contained within at least 50% of the model were then identified.

Table 15 shows the biomarker frequency in predicting the invasion assessment. The performance of the top ranking model was similar. In addition, the number of biomarkers in the top ranking model varied. To solve this problem, which appears to be related to model size, we considered the top 1% of the models sorted by test AUC. We considered distributions obtained with several different population assumptions, such as when the intermediate core GS was excluded from the analysis, or included a slow or chronic score, or included a high score. In the final analysis, we concluded that the 8 biomarker model provides the best trade-off between performance and complexity in this experimental data set.
Biomarker ranking for lethality by thorough search of multi-marker models

  The same model building approach as biomarker ranking for predicting lethality was followed. Table 16 shows biomarker usage frequency (top 5%) for lethality.

Integrating results into the final biomarker set The final set of 12 selected biomarkers must reflect its biological significance as assessed in univariate and multivariate analysis of patient sample measurements. there were. What complicates and suppresses the final selection has taken into account the technical limitations of the specific MAbs available for testing. The selected final biomarker set is shown in FIG.

Supplementary Appendix Results To develop a locked down model for clinical use, the 12 markers identified in this study were advanced to another independent study of prostate cancer FFPE biopsy samples (manuscript was Submitted). In this new study, we identified the best marker subset of these 12 markers and were obtained including the following biomarkers: SMAD4, FUS, CUL2, YBX1, DERL1, PDSS2, HSPA9 and pS6 8- The marker model was locked down. For completeness, we analyzed this set of markers in TMA samples in this study with the understanding that the TMA cohort helps the marker selection process. We again used the same patient partition and tested with both L TMA and H TMA samples after training in L TMA. We analyzed 268 patients with 40 disease-related death events. The test AUC obtained based on the LTMA for the prediction of aggressive disease is 0.64 (95% CI: 0.56-0.71), the test odds ratio for aggressive disease is 13 / change in risk score The unit was 95% CI: 2.3 to 341. The test hazard ratio for predicting lethal outcome was 14 / unit of change in risk score (95% CI: 1.3-393). To confirm the possibility of generalization under sampling error, a model derived from L TMA training was also tested in Test H TMA, and the results were consistent for both indications. H TMA test AUC is 0.70 (95% CI: 0.62 to 0.78), odds ratio of aggressive disease is 46 / unit of change in risk score (95% CI: 5.6 to 1290) there were. The test hazard ratio of HTMA for predicting lethal outcome was 19 / unit of change in risk score (95% CI: 1.4-620).

Materials and Methods Quantitative Multiplex Immunofluorescence (QMIF) Staining Procedure QMIF consisted of two initial blocking steps followed by four MAb incubation steps (with appropriate washing between steps). Blocking consisted of a biotin blocking step followed by treatment with Snipper reagent (Biocare Medical) (according to manufacturer's instructions). The first MAb incubation step comprises a mixture of anti-biomarker 1 mouse MAb and anti-biomarker 2 rabbit MAb, followed by a second mixture comprising anti-mouse IgG Fab-fluorescein isothiocyanate (FITC) and anti-rabbit IgG Fab-biotin. It consisted of a process. The third “visualization” step consists of anti-FITC MAb-Alexa 568, streptavidin-Alexa 633 and MAb against epithelium (anti-CK8-Alexa 488 and anti-CK18-Alexa 488) and MAb against basal epithelium (anti-CK5-Alexa 555), respectively. And anti-Trim29-Alexa 555) mixture. The final fourth step included a brief incubation with 4 ′, 6-diamidino-2-phenylindole (DAPI) for nuclear staining. After the last washing, the slide was placed on Prolong Gold ^™ (Life Technologies) and then covered with a cover glass. Slides were maintained at -20 ° C unchanged before and after imaging.

Quality Evaluation of FFPE Tissue Blocks 5 μm sections from each FFPE block were manually stained with anti-phosphoSTAT3 (T705) rabbit MAb, anti-STAT3 mouse MAb and region of interest markers as described above. The slides were visually inspected under a fluorescence microscope. Based on staining intensity and autofluorescence, sections and their corresponding FFPE blocks were classified into four quality categories.

Image Acquisition Two Vectra Intelligent Slide Analysis Systems (PerkinElmer) were used as described elsewhere for automatic image acquisition. DAPI, FITC, tetramethylrhodamine isothiocyanate (TRITC) and Cy5 long pass filter cubes were optimized for maximum multiplex capability. Vectra 2.0 and Nuance 2.0 software packages (PerkinElmer) were used for automatic image acquisition and spectral library development, respectively.

  The TMA acquisition protocol was performed according to the manufacturer's instructions (PerkinElmer) in automatic mode. Two 20 × fields per core were imaged using a multispectral acquisition protocol including continuous exposure with DAPI, FITC, TRITC and Cy5 filters. For maximum reproducibility, images of each TMA slide were acquired after adjusting the light source intensity with the aid of X-Cite Optical Power Measurement System (Lumen Dynamics). All slides containing the same antibody combination used the same exposure time. A set of TMA slides stained with the same antibody combination was imaged with the same Vectra microscope.

  Spectral profiles of each fluorescent dye as well as FFPE prostate tissue autofluorescence were generated. Interestingly, two types of autofluorescence were observed in FFPE prostate tissue. While typical autofluorescence signals were common to both benign and tumor tissues, the atypical “bright” type autofluorescence was mostly specific to light granules present in epithelial cells of benign tissues It was the target. A spectral library containing a combination of the two spectral profiles was used to separate or “unmix” individual dye signals from the autofluorescent background.

FFPE Cell Line Control Selected cell lines were cultured under standard conditions, with or without pre-treatment as indicated (Table 18). Cells were washed with phosphate buffered saline (PBS) and fixed directly on the plate with 10% formalin for 5 minutes, then scraped and collected in PBS, and fixation was continued for 1 hour at room temperature. Cells were washed twice with PBS, resuspended in Histogel (Thermo Scientific) at 70 ° C. and quickly spun down in a 1.5 ml microcentrifuge tube to form a concentrated cell-Histogel pellet. This pellet was then embedded in paraffin to a standard paraffin block, which served as a donor block for TMA construction. DU145 cells with inducible knockdown of CCND1 and SMAD4 were established using the “Tet-one” system (Clontech) according to the manufacturer's instructions.

Antibody Specificity Assays Several MAbs, including anti-ACTN1, anti-CUL2, anti-dellin 1, anti-FUS, anti-PDSS2, anti-SMAD2, anti-VDAC1, anti-YBX1 and anti-HSPA9, target specific knockdown cells and control cells. Validated by Western blotting (WB) and immunohistochemistry (IHC) assays (Figure 37). Details of the small interfering RNA (siRNA) sequences and host cell lines are shown in Table 19. Cells were seeded in 12-well plates and transfected with 25 nM siRNA and DharmaFect transfection reagent (Thermo Scientific Dharmacon); mock transfection included only transfection reagent. Cells transfected with two non-targeting sequences were also included as controls.

  For the WB assay, transfected cells were harvested at 72 hours and lysed using Pierce RIPA buffer (Thermo Scientific) supplemented with Halt protease inhibitor cocktail (Thermo Scientific). The protein concentration was measured using Pierce BCA reagent (Thermo Scientific). Samples were adjusted to equal protein concentrations, then mixed with sample buffer (Boston BioProducts) and run on precast Criterion TGX 4-15% SDS-PAGE gel (Bio-Rad). Samples were transferred onto PVDF or nitrocellulose membranes using an IBlot apparatus (Life Technologies), immunoblotted overnight at 4 ° C. with antibodies, and then incubated with secondary mice or rabbit MAbs (Sigma Aldrich). Blots were developed using SuperSignal West Femto reagent (Thermo Scientific) and visualized by exposure with FluorChem Q system (Protein Simple).

  In the IHC assay, cells cultured on a cover glass in a 12-well plate were fixed with methanol for 20 minutes on ice 72 hours after transfection. This was followed by permeabilization for 10 minutes with 0.2% Triton X-100 on ice. For subsequent IHC assays, UltraVision LP Detection System HRP Polymer / DAB Plus Chromogen Kit (Thermo Scientific) was used according to the manufacturer's instructions.

  SMAD4 antibodies were verified by WB and IHC assays of SMAD4 positive cell line PC3 and SMAD4 negative cell line BxPC3. Phospho-S6 antibody was verified by WB and IHC of untreated DU145 cells and LY294002 treated DU145 cells.

Cell Proliferation Assay HeLa cells were monovalently transfected with two non-targeting siRNAs as well as si9-11 specific for HSPA9 (see Table 19 for siRNA sequence details). Forty-eight hours after transfection, cells were re-plated in 96-well plates and seeded in triplicate at 1000 cells / well. Cell growth was monitored at 0, 24, 72 and 120 hours after re-plating using the CellTiter-Glo® Luminescent Cell Viability Kit (Promega) according to the manufacturer's instructions.

Clonogenicity assay 48 hours after transfection, HeLa cells were re-plated at 500 cells / well in 6-well plates containing 2 ml of cell culture medium. Seven days after plating, the cells were fixed using Crystal Violet Solution (Sigma). Images of each well were acquired using AlphaView software in the FluorChem Q system (Protein Simple) and processed using ImageJ software.

Cell viability assay HeLa cells were harvested 120 hours after transfection. Cells were collected using trypsin. Cell pellets from each well of a 12 well plate were suspended in 500μ cell culture medium. Cell suspension (95 μl) was mixed with 5 μ Solution 5 (VB-48 / PI / AO) and 30 μ mix was loaded onto NC-Slide A2 (both from ChemoMetec). Difference cell viability was measured by NucleoCounter NC-3000 ^™ (ChemoMetec) according to manufacturer's instructions.

Caspase assay HeLa cells were harvested with trypsin 120 hours after siRNA transfection. The cells were suspended at 2 × 10 ⁶ cells / ml. A 93 μ aliquot of the cell suspension was mixed with 5 μl of diluted FLICA reagent (ImmunoChemistry Technologies) and 2 μ of Hoechst 33342 (Life Technologies). The mixture was incubated at 37 ° C. for 1 hour. HeLa cells were washed twice with 1 × Apoptosis Buffer (ImmunoChemistry Technologies). The cell pellet was suspended in 100 μl of 1 × Apoptosis Buffer and 2 μ of propidium iodide. A 30 μl aliquot of the mixture was loaded onto NC-Slide A2 and read using NucleoCounter NC-3000 software for caspase assays. Cells positive for FLICA staining were counted as apoptotic cells.

Identification of HSPA9 (Mortarin) Preparative immunoprecipitation was performed to identify the Leica “anti-DCC” antibody target (FIG. 37). Ten p100 plates of confluent A549 cells were collected using 5 ml of RIPA buffer (Thermo Scientific) plus protease inhibitors. The cell lysate was spun down at 14,000 rpm for 5 minutes; the supernatant was heated at 80 ° C. for 5 minutes, then cooled on ice and spun down again at 14,000 rpm for 5 minutes. The supernatant was collected and 50 μ Protein A / G beads (Thermo Scientific) pre-bound with 2 μg of “anti-DCC” antibody were added followed by incubation at 4 ° C. for 2 hours with shaking. The beads were washed 3 times with TBS + 1% Triton X100 and boiled with 30 μl of 1 × SDS-PAGE loading buffer. The supernatant was loaded on a 10% SDS-PAGE gel and separated under standard SDS-PAGE conditions. This gel was stained with a silver staining kit for mass spectrometry (Thermo Scientific); specific bands were excised, digested with trypsin and subjected to MS / MS sequencing mass spectrometry at Taplin Mass Spectrometry Facility (Harvard Medical School). Provided. The identified peptides were aligned with the Human Protein reference database. The identified protein HSPA9 was further verified as described.

Example 7: Clinical Validation Summary of Proteomics In Situ Biopsy Trials for Distinguishing Prognosis and Prognosis of Prostate Cancer The aggressiveness and appropriate treatment of prostate cancer is determined according to the sampling of the biopsy material. Currently used clinical and pathological parameters are inadequate for accurate risk prediction and are not only over-treated, but also lose the opportunity for curative treatment.

  An 8-biomarker proteomics assay for intact tissue biopsy to predict prostate pathology was defined in a study of 381 patient biopsies with matched prostatectomy specimens, followed by 276 patient cases Validated in a blinded manner. The possibility of distinguishing pathologically “good prognosis” and “poor prognosis” disease profiles based on prostatectomy was examined relative to current standard care (SOC) for risk classification.

  The validation study was satisfactory to separate the two predetermined endpoints with good prognosis from poor prognosis (AUC, 0.68, P <0.0001, odds ratio = 20.9). Patient categories with good prognosis (risk score ≦ 0.33) and poor prognosis (risk score> 0.80) were defined based on “false negative” and “false positive” rates of 10% and 5%, respectively. With a risk score ≦ 0.33, the predicted values for patients with good prognosis in the very low and low risk NCCN and low risk D'Amico groups are 95%, 81.5% and 87.2%, respectively It was higher than that of the group (80.3%, 63.8% and 70.6%, respectively). The predictive value for patients with poor prognosis was 76.9% with a risk score> 0.8 in all risk groups. In all risk groups, a high risk score correlated with a lower frequency of cases with good prognosis. The net reclassification improvement at NCCN was 0.34 (P <0.00001) and at D′ Amico was 0.24 (P = 0.0001).

  This 8-biomarker test can provide independent, independent and complementary information to the SOC risk stratification system information and assist in clinical decisions at the time of biopsy.

Introduction In 2014, an estimated 233,000 cases were newly diagnosed with prostate cancer in the United States ¹ . Most patients are patients with early stage clinically localized disease ^1-5 . Considering the concerns regarding significant heterogeneity of prostate cancer and its over-treatment, ^6-8 , post-biopsy and prior to definitive treatment, inferior cases with good prognosis are more aggressive cases with poor survival. It is important to identify ⁹ . The pathological assessment of the tissue obtained by needle biopsy is essential both to confirm the prostate cancer diagnosis and to determine the patient's risk category ¹⁰ . Several classification systems that combine available clinical and pathological parameters have been developed ^9,11 . However, not all classification systems are perfect and none were designed to confirm individual risk scores ^12-14 .

Approximately 25-30% of patients considered to have low-risk disease at the time of diagnosis will subsequently increase in malignancy of the tumor pathology ^14-16 . In fact, a significant proportion of patients are graded up or down from the initial “biopsy” Gleason score to a more accurate “surgical” or pathological Gleason score after analysis of the radical prostatectomy tissue ¹⁶ . Such correction may reflect an initial biopsy sampling error ¹⁷ or a discrepancy ¹⁸ in the pathologist's tumor classification, both of which may contribute to over-treatment or under-treatment of the disease ⁷ . In particular, there are concerns about over- or under-determination of Gleason pattern 4 in needle biopsy samples ^{19 16,20,21} , ^whether the cancer is organ-localized or ultimately organ-localized with the potential for metastasis, There is a continuing need to be able to determine that a patient in a biopsy material has a low to moderate grade disease.

The identification of genetic markers that provide clinical risk prognostic information is progressing, one example being the expression of a set of cell cycle progression genes that are used to predict mortality risk ^22-25 . In addition, in situ identification of protein biomarkers that can be measured in the most aggressive tumor area, even with very few cancer cells in the context of tumor heterogeneity ^26-28 . Using a quantitative multiplex proteomics in situ imaging (QMPI) approach, the inventors were tailored to be resistant to sampling errors in independent large clinical trials, and aggressive prostate pathology and lethality Twelve biopsy biomarker candidates that predicted both outcomes were identified (see Example 6 below and supplementary appendix).

Here, in two separate biopsy clinical trials (each with a matched annotated prostatectomy specimen), development of a model of 8 biomarker indices derived from these 12 markers and subsequent blindness Report verification. The first study was designed to lock down the biomarker indicator model and QMPI assay (ProMark ^™ ) as defined by logistic regression (training-test) analysis and to obtain a risk score for possible disease aggressiveness. . A blinded clinical validation study evaluated the potential for biopsy assays to predict clinically relevant dichotomized endpoints, good prognosis and poor prognosis, at prostatectomy. Compare the information gained in the assay and risk score with the two risk stratification systems, the D'Amico system and the NCCN guideline categories ⁹ , ¹¹ , for further accuracy in predicting the prognosis of individual patients as a potential aid in decision making Considering the possibility of

Methods The QMPI approach for in situ measurement of proteins is as described in the supplementary appendix below.

Clinical model design studies and assay lockdowns to define the best marker subset index from 12 previously identified biomarker candidates that have been shown to correlate with both aggressive aggressiveness and lethal outcome of the prostate We devised a developmental study of a non-interventional retrospective clinical model using tissue samples at biopsy. The purpose of this study was to investigate the pathology of prostate with surgical Gleason 3 + 3 and ≦ T3a (“GS6”) and prostate with surgical Gleason ≧ 3 + 4 or non-local> T3a, N or M (“non-GS6”) ^29,30 was to define a model that could distinguish pathology based on studies showing that tumors with surgical Gleason 3 + 3 do not metastasize at prostatectomy ^29,30 . The study protocol was approved by the Institutional Review Board (IRB) and patient consent was obtained or withdrawn accordingly.

  To develop a robust assay, recruitment was conducted at other institutions that are typical US patient cohorts: Urology Austin, Chespeake Urology Associates, Cleveland Clinic, Michigan Urology and Folio Biosciences. Biopsy sample inclusion / exclusion criteria were compatible with what would be appropriate during routine clinical use of the assay (supplemental appendix). Patients with biopsy Gleason ≧ 4 + 3 were excluded except for a limited number of biopsies whose classification by two specialist pathologists was inconsistent with both 3 + 4 and 4 + 3. Annotations were needed that included information on biopsy fit and prostatectomy pathology reports. All samples were blinded during laboratory processing.

The biomarker index was optimized as a logistic regression model to estimate the probability of “non-GS6” determined by bootstrap analysis of independent training and test sets. The model is characterized by the receiver operating characteristic (ROC) area under the curve (AUC), increasing the value of the Akaike Information Criterion (AIC) ³¹ , decreasing the value of AUC in the training case set, and the AUC in the test set Sorted by decreasing value. The marker usage was then examined in the highest ranking 10% model and the biomarker set was finalized. A consecutive number of risk scores between 0 and 1 were computerized to estimate the likelihood of a “non-GS6” condition. Sensitivity analysis was performed to confirm the defined lockdown assay.

Clinical validation trials Non-interventional, blinded, prospectively designed and retrospectively collected clinical studies, performing 8 biomarker biopsy assays alone in predicting prostate pathology, and current patient risk categorization It was verified by comparing with standard medical care (SOC). The cohort consisted of biopsy samples with matched prostatectomy annotations of patients managed at the University of Montreal (Canada). Consent criteria and IRB approval measures were as in clinical development studies. Inclusion criteria were focused Gleason score 3 + 3 or 3 + 4 biopsies (including 3 + 4 and 4 + 3 mismatched biopsies by 2 specialist pathologists) as well as pathological TNM progression classification, PSA Level and Gleason score matched prostatectomy. The performance of the assay was assessed using a diagnostic risk score ROC and the corresponding AUC.

Prostate pathology The two co-primary endpoints were validated by biopsy assay derived risk score assessed by AUC:
1. “Good prognosis” pathology—surgical Gleason ≦ 3 + 4 and organ-localized disease (≦ T2) “poor prognosis” pathology—surgical Gleason ≧ 4 + 3 or non-organ-localized disease (T3a, T3b, N or M)
And 2. “GS6” pathology—surgical Gleason 3 + 3 and localized disease (≦ T3a) “non-GS6” pathology—surgical Gleason ≧ 3 + 4 or nonlocalized disease (T3b, N or M)

Good prognosis and poor prognosis (contrast) were selected by validation studies for final patient categorization. This is due to the growing recognition that organ-confined diseases with minimal Gleason 4 patterns remain probably harmless and have a significantly better long-term prognosis than high-grade (predominant Gleason 4 patterns) or non-organ-confined diseases Reflecting ^{30, 32, 33} .

The secondary analysis included the highest quartile to lowest quartile odds ratio (OR) of risk scores and OR (point estimation) of consecutive steps. We compared the risk outcome of our diagnostic tests with SOC risk classification categories ⁹ , ¹¹ defined by D'Amico and NCCN, using positive predictive value (PPV). The definition and statistical analysis of net reclassification improvement (NRI) was performed as described in Pencina ³⁴ .

  See supplemental appendix for statistical plans for both clinical studies.

Results The tumor characteristics of 381 patients included in the clinical model design study and the assay lockdown model development study are shown in Table 20. 39A-C show the model optimization process. FIG. 39A shows the univariate OR associated with the evaluated biomarker. Model performance was evaluated and several high performance models were identified, such as test AUC 0.79 (95% confidence interval [CI], 0.72-0.84). FIG. 39B shows biomarker frequencies obtained in all models using up to 8 biomarkers. The obtained lockdown index is shown in FIG. 39C.

Clinical validation studies Table 20 summarizes the tumor characteristics of 276 samples of clinical validation studies. As shown in Table 21, in this study, two co-primary endpoints were met and the assay was validated for both endpoints (good prognosis: AUC, 0.68 [95% CI, 0.61-0 P <0.0001; OR of risk score, 20.9 / unit of change; “GS6” pathology: AUC, 0.65 [95% CI, 0.58 to 0.72]; P <0. 0001: OR of risk score, 12.6 / change unit). Further details are shown in FIGS.

  We had enough annotations to classify 256 cases according to the NCCN and D'Amico criteria. The performance of the biomarker indicator assay in this cohort for favorable prognosis is shown in FIG. Performance was similar for all cohorts (AUC, 0.69 [95% CI, 0.63 to 0.76]; P <0.0001; OR of risk score, 26.2 / variation unit).

  FIG. 40 shows the sensitivity and specificity associated with risk scores as an aid to prognosis for good / poor prognosis and the distribution of risk scores in the NCCN and D'Amico categories. FIG. 40A shows an example of a good prognosis category identified based on the molecular index in this test population. A threshold of 0.33 for a good prognosis category yielded a sensitivity of 90% (95% CI, 82% to 94%) (P [risk score> 0.33 | poor prognosis]), which is a prognosis The false negative rate is reduced to 10% (95% CI, 6% to 18%) in patients with poor pathology. Similarly, in FIG. 40B, a poor prognosis category with 95% (95% CI, 90% to 98%) specificity (P [risk score ≦ 0.80 | good prognosis]) was identified in this study population. This reduces the false positive rate to 5% (95% CI, 2% to 10%) in patients with good prognosis.

We evaluated the predictive value of the risk score and compared it to those of the NCCN and D'Amico risk categories (Table 22). The PPV with a good prognosis identified with a risk score of ≦ 0.33 was 83.6% (specificity, 90%). Conversely, at a risk score> 0.80, 23.1% of patients had a good prognosis disease (ie, 76.9% had a poor prognosis disease). Based on this test population, this translates to 39% of patients with a risk score of ≦ 0.33 or> 0.8, of which 81% are correctly identified.

  We further examined the distribution in each NCCN category of patients with good prognosis according to our risk score (Figure 40C and Table 22). Using a risk score of ≦ 0.33, the PPV for good prognosis was 75% for NCCN intermediate risk, 81.5% for NCCN low risk, and 95% for NCCN very low risk (FIG. 40C and table) 22). This is in contrast to the PPV obtained in the NCCN risk category alone (40.9% for intermediate risk, 63.8% for low risk and 80.3% for very low risk) (Table 22). . Thus, this risk score provided further information about individual patients for the NCCN risk category. As shown in FIG. 40D, an increase in risk score correlated with a decrease in good prognosis case frequency in each NCCN category. Similar results were obtained when compared to the D'Amico category (Figures 40E and 40F).

  To confirm the benefit of the risk score in relation to SOC, we performed an NRI analysis for the good prognosis and poor prognosis categories compared to NCCN and D'Amico. Using the basic data shown in Table 22, we found that the NCRI NRI was 0.34 (P <0.00001; 95% CI, 0.20-0.48), and 0 for D'Amico. .24 (P = 0.0001; 95% CI, 0.12-0.35; see FIG. 44). This shows that this test provides additional discriminatory power to that shown by the SOC classification system alone.

Discussion We report here the results of two clinical studies conducted on prostate cancer biopsy samples using matched prostatectomy samples: development studies and blinded validation studies. These studies demonstrate the accuracy and validity of a novel proteomic multi-biomarker assay that can be used to predict the presence of high-risk features in the prostate and the possibility of extraprostatic expansion and metastasis at biopsy. In our first model building trial, the optimal 8 biomarker index was determined from 12 candidate biomarkers previously shown to predict tumor aggressiveness and lethality. The 8 biomarker indices as well as the individual risk scores obtained by this study are based on logistic regression analysis of the prediction of “non-GS6” prostate pathology (surgical Gleason score ≧ 3 + 4 or nonlocality> T3a, N, M) Stipulated.

  Secondly, blinded clinical validation studies include the following: “GS6” pathology (defined as surgical Gleason 3 + 3 = 6 and ≦ T3a) and “good prognosis” pathology (pathology of organ-localized prostate (surgical) Met their two co-primary clinical endpoints predicting independent prostate pathology of clinical and pathological parameters (Gleason 3 + 3 or 3 + 4; defined as ≦ T2), and our risk score Adds personalized information that is discriminative and complementary to the SOC risk stratification.

In recent studies, the long-term survival of patients with organ-localized Gleason 3 + 4 disease is significantly better than patients with non-organ-localized disease or tumors with dominant Gleason pattern 4 or higher ^{19,32 ,} long-term outcome by deferred treatment for ³³ and former group has been shown ^35-37 to not change significantly. Currently, most risk stratification systems do not distinguish between Gleason 7 biopsies, and patients who are typically considered candidates for active surveillance typically contain only a biopsy Gleason score ≦ 6 ^3,9 belonging to the "low risk" or "low risk" group. However, when compared to surgical Gleason, around 25% of Gleason classification 3 + 4 biopsies are “graded down” and a similar percentage of Gleason classification 3 + 3 biopsies are “graded up”. This is due to biopsy sampling errors and pathologist inconsistencies ^{16 20} . On this basis, Gleason grade 3 + 3 and 3 + 4 biopsies need high ²¹ molecular evidence-based tests for the ^material, the present inventors have developed a test of the present inventors for this indication. Cases with good prognosis (surgical Gleason 3 + 3 or 3 + 4, organ-localized [≦ T2] tumor) and poor prognosis (extraprostatic extension [T3a], seminal vesicle invasion [T3b], lymph node or distant metastasis or 4 or more We have developed a good prognostic endpoint to discriminate between high dominant Gleason patterns).

  Our study shows that with a test risk score ≦ 0.33, the predicted values for identifying patients with good prognosis in the very low risk NCCN and low risk NCCN and low risk D'Amico groups are respectively 95%, 81.5% and 87.2%, indicating higher values than those obtained with these risk groups alone. In addition, this study will undoubtedly treat patients with poor prognosis that are not suitable for active monitoring with a predictive value of 76.9% for both risk stratification systems with a risk score> 0.8 in all risk groups. It can also be identified with high reliability. The significance of patient stratification based on this study for individual patients is indicated by the fact that an increase in test risk score is correlated with a decrease in the frequency of good prognosis cases observed in all risk stratification groups. A further measure of information obtained by the risk score compared to SOC was obtained by NRI analysis. We found that the NRI of NCCN was 0.34 (P <0.00001; 95% CI, 0.20 to 0.48), and that of D'Amico was 0.24 (P = 0.0001; 95% CI). , 0.12-0.35). Of patients with good prognosis and poor prognosis, 78% and 76%, respectively, were correctly adjusted to lower and higher risks than apparent in the NCCN risk group itself (FIG. 44).

  In embodiments, our risk score is generated based on quantitative measurements of eight biomarkers in intact tissues using a multiplex proteomic imaging platform (supplemental appendix). This approach has several potential advantages over gene expression-based tests that homogenize tissues prior to analysis. First, this study is robust to the diversity of benign tissue to tumor tissue ratios. This is because it does not interfere with marker measurement in intact cancer cells. In addition, this test allows the incorporation of molecular and morphological information and only a few cancer cells are required.

The eight biomarkers in our model constitute a subset of twelve biomarker candidates identified as predicting both aggressive and lethal outcomes despite tissue sampling errors. This indicates that the pathology endpoint used in this study is also associated with long-term outcome, as ^{reported 29,32,33} .

In conclusion, our results show the utility of this clinical biomarker biopsy test for personalized prognosis of prostate cancer and its impact on treatment options. The ability to improve individual patient information compared to SOC, which currently has limited prognostic capabilities, is a useful aid in clinical decisions.

Material Supplements FIGS. 41A-E, 42A-C, 43A-C, 44A-B and 45A-C provide additional information and are described in the brief description of the above drawings.

Method for Quantitative Multiplex Proteomic Imaging (QMPI) Formalin-fixed paraffin-embedded (FFPE) prostate cancer biopsy tissue slides for intact tissue with integrated morphological object recognition and molecular biomarker measurements of tumor epithelium Analyzes at the individual slide level using a quantitative multiplex proteomic imaging (QMPI) platform. Antibody validation, staining protocol, image acquisition, image analysis and inter-experimental control are described below.

Assay Description and Biomarker-Antibody Validation The assay was performed using four slides outlined in the staining protocol illustrated in FIG.

  Four combinations of 3 (Triplex) biomarkers were used, respectively: A) PLAG1, SMAD2, ACTN1; B) VDAC1, FUS, SMAD4; C) pS6, YBX1, DERL1; D) PDSS2, CUL2, DCC. Each primary antibody used was verified for specificity and was found to have insufficient specificity for PLAG1; it was therefore excluded from the potential index. Each triplex assay consisted of an initial blocking step followed by 5 consecutive incubation steps (with appropriate washing between steps).

  1) Incubation with a mixture of anti-biomarker 2 (rabbit monoclonal antibody [MAb]) and anti-biomarker 3 (mouse MAb)

  2) Incubation of a mixture of Zenon anti-mouse IgG Fab-horseradish peroxidase (HRP) and Zenon anti-rabbit IgG Fab-biotin

  3) Incubation with anti-biomarker 1 MAb conjugated to FITC

  4) Visualization using a mixture of anti-FITC MAb-Alexa 568, streptavidin-Alexa 633, anti-HRP-Alexa 647, anti-CK8-Alexa 488, anti-CK18-Alexa 488, anti-CK5-Alexa 555 and anti-Trim29-Alexa 555 Process

  5) Short incubation with DAPI for nuclear staining

  After the final washing, the slide was placed in Prolong Gold (Life Technologies), covered with a cover glass, and the slide was stored at −20 ° C. overnight, and then image acquisition was performed.

Slide processing and staining protocol The majority of the slide processing and staining process was automated to ensure maximum reproducibility. Sections were first deparaffinized using StainMate (Thermo Scientific) in xylene / grade alcohol. Antigen activation was performed using a Lab Vision PT module (Thermo Scientific) for 45 minutes at 95 ° C. using 0.05% cytosolanic acid anhydrous solution. Slides were stained with Autostainer 360 or 720 (Thermo Scientific) using the assay format described above. Samples in the case of biopsy material were stained with a batch of 25 slides / Autostainer, and one cell line tissue microarray (TMA) was a control slide (see below) for each triplex assay format.

Image Acquisition In each triplex assay, one specific Vecta Intelligent Slide Analysis System (200-slide volume) is used for quantitative multiplex immunofluorescence image acquisition to maximize spectral resolution and minimize inter-fluorophore bleed. Optimized DAPI, FITC, TRITC and Cy5 long pass filter cubes were used to allow for thru. To minimize the difference, the light intensity in each system was calibrated before each run using the X-Cite Optical Power Measurement System (Lumen Dynamics). Vectra 2.0, Inform 1.3 and Nuance 2.0 software (PerkinElmer) were used for image acquisition, tissue search algorithm creation and spectral library development, respectively.

  In the image acquisition process, first, an image of the entire slide was acquired with a 4 × monochrome DAPI filtered image mosaic. The initial tissue search algorithm included in the image acquisition protocol was then used to determine the tissue location, which was then subjected to image acquisition again, using both 4 × DAPI and 4 × FITC monochrome filters. . The final tissue search algorithm included in this protocol was then applied to ensure that an image containing a sufficient amount of tissue in all 20x fields was acquired (FIG. 45B).

  The algorithm included in the image acquisition protocol limited data collection to a 20x field of view containing a sufficient amount of tissue. The multispectral acquisition protocol used for the assay had continuous exposure of DAPI, FITC, TRITC and Cy5 filters. Once image acquisition was complete, the image cube was automatically stored on the server for subsequent automatic unmixing into individual channels and processing by the Definiens software.

Inputs for Image Analysis and Risk Scoring Model We have developed an image analysis algorithm for tumor identification and biomarker quantification using Defieniens Developer XD (Definiens AG, Munich, Germany). Software was used to delineate the malignant and benign epithelial areas of the biopsy tissue, allowing for the measurement of marker strength exclusively in the malignant areas. For each biopsy sample, several 20 × image fields were scanned using CRi Vectra (PerkinElmer) and saved as multispectral image files. In order to acquire an image of the entire tissue sample, as many as 140 individual fields were scanned for a given slide. Triplicate 11 different FFPE cell lines and duplicate 2 prostatectomy tissue samples were used as controls in separate quality control slide arrays, with 2 for each 1.0 mm quality control cell line or tissue core. A 20 × image field was scanned (ie, a total of 6 images for each cell line control and 4 images for each tissue control). The Vectra multispectral image file was first converted to a multi-layer TIF format using inForm (PerkinElmer) and a customized spectral library, and then converted to a single layer TIFF file using BioFormats (OME). Customized import algorithm to load a single layer TIFF file into the Definiens workspace, and for each biopsy sample and each quality control, all image field view TIFF files are loaded as "maps" and analyzed in a single "scene" Was imported using.

  Automatic adaptive binarization was used to define a fluorescence intensity cutoff for tissue segmentation at each individual tissue sample in our image analysis algorithm. Cell line control cores were automatically identified from prostatectomy tissue cores based on predetermined core coordinates in the Definiens algorithm on quality control slides. Biopsy and tissue core samples were segmented using fluorescent epithelial and basal cell markers with DAPI for classification into epithelial cells, basal cells and stroma, and further partitioned into cytoplasm and nucleus. Individual gland areas were classified as malignant or benign in combination with object-related features, such as gland thickness, based on relevant features between basal cells and adjacent epithelial structures. There was no epithelial marker in all cell lines and therefore cell line controls were segmented using autofluorescent channels in tissue versus background. Fields with artifact staining, insufficient epithelial tissue and defocused images were removed by a rigorous multi-parameter quality control algorithm.

  The intensity of epithelial markers, DAPI, ACTN, VDAC and DERL1 were quantified as quality control measures in malignant and non-malignant epithelial areas. Biomarker values were also measured in the cytoplasm, nucleus and whole cells of malignant and non-malignant epithelial areas. The average biomarker pixel intensity for each cell component compartment was averaged in each individual map with acceptable quality parameters, and map specific values were exported for bioinformatics analysis. A weighted average was calculated from the appropriate values to obtain a single intensity for each marker in the tissue sample; a 20x field with a mean intensity value of the 40th to 90th percentile on the slide or a 20x field containing a large tumor area Was considered preferred. This provided input for the risk score model.

Inter-experimental control: quality control procedure A cell line control was used as a batch control. All samples in the case of received biopsies were also subjected to a multi-step quality control procedure, as a means of sample inclusion or exclusion for clinical studies. Raw slides with sections were examined for staining and the presence of dye by visual and fluorescent microscopy. Samples with significant amounts of fluorescent dye in the biopsy tissue were excluded from further analysis. This is because it can be in clinical pathology practice. Next, one of the sample slides for each biopsy was manually stained with ACTN1, CK8 / 18-Alexa 488 and CK5 / Trim29. The stained slides were examined manually; if the tissue is small or fragmented, there is little tumor tissue, or the staining with any of the three markers above is insufficient, the sample is Quality control failed.

  After multiplex immunofluorescence staining, all 20x images were examined manually, and fields containing sham / non-prostate tissue (eg, intestinal tissue) were excluded from further analysis. When the malignant tissue and benign tissue were separated by image analysis, cases where the benign region or tumor region was insufficient were excluded. Also excluded was the case where the ACTN1, DERL1 or VDAC level was below a predetermined minimum value.

Staining control development and application: Cell line control 30 cell lines were stained with each marker used in this study, 11 of which were based on minimal range, signal intensity and differences Selected for control staining.

  Cell lines were uniformly cultured to 70% -80% confluence in defined media and fixed on plates using formalin. Cells were scraped and spun down, and a cell disc was prepared from the cell / histgel suspension of cell pellets, which was embedded in paraffin. This pellet was used to make a TMA block for use in reproducibility studies, for use in master mix validation and as a control slide during routine sample staining.

  One of the sections / slides from cell line TMA was processed with each batch of biopsy slides. Staining, image acquisition and data extraction and analysis were performed in exactly the same manner as described above for the individual triplex assay formats.

Clinical Trials: Statistical Plans After locking and recording the statistical analysis plan (SAP) and communicating with an external biostatistics department, clinical study data became available for analysis in validation trials. According to SAP, all P values of bilinear outcomes are reported after multiplying by 2 to reflect Bonferroni correlation. AUC CI and P values were estimated using the binomial exact test, whereas, AUC standard error was determined using the method described in DeLong et al 1988 ^1. OR by logistic regression was included in comparison with SAP and standard care using binomial accuracy CI for positive predictive value, sensitivity and specificity. In other cases, statistical personnel who were not involved in assay development performed statistical analysis.
References
1． DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837-45.

Claims (229)

A method of assessment of a cancer sample from a patient, e.g. a prostate tumor sample, e.g. a computer-implemented method or an automated method, comprising:
1, 2, 3, 4, 5, 6, 7 or 8 tumor markers of FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set), or DNA or mRNA of said tumor marker Identifying the level of, e.g., amount or expression level,
Thereby evaluating the tumor sample. 2. The method of claim 1, comprising obtaining a tumor marker signal, for example, directly or indirectly. The method of claim 2, comprising obtaining the signal directly. The method of claim 1, comprising obtaining the sample directly or indirectly. A cancer sample; and 1, 2, 3, 4, 5, 6, 7 or 8 tumor markers of FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set) or said tumor marker A reaction mixture containing a detection reagent for DNA or mRNA. 6. The reaction mixture of claim 5, wherein the cancer sample comprises a plurality of portions, such as slices or aliquots. The first portion of the cancer sample includes a detection reagent for a first marker that is not all of the marker, and the second portion of the cancer sample does not include a detection reagent for the first marker. The reaction mixture according to claim 5, comprising a detection reagent for one of the markers. A method of evaluation of a sample from a patient, e.g. a tissue sample, e.g. a cancer sample, e.g. a prostate tumor sample, e.g. a computer-implemented method or an automated method,
(A) identifying the level of a phenotypic marker of a first region, such as a first tumor marker, in the region of interest (ROI) from the sample;
Thereby evaluating the sample. 9. The method of claim 8, wherein the sample is a cancer sample. The method according to claim 8, wherein the sample contains cells derived from a solid tumor. The method according to claim 8, wherein the sample contains cells derived from a liquid tumor. The method of claim 8, wherein the ROI is defined or selected by morphological features. The ROI is defined or selected by manual or automated means and physical separation of the ROI from other cells or materials, eg, by ROI from other tissues, eg, non-cancerous cells, eg, incision of a cancerous region The method according to claim 8. . 9. The method of claim 8, wherein the ROI is defined or selected by non-morphological features, such as ROI markers. 9. The method of claim 8, wherein the ROI is identified or selected by cell sorting by including ROI markers. 9. The method of claim 8, wherein the ROI is identified or selected by a combination of morphological and non-morphological selection. The level of a first region phenotypic marker, eg, a first tumor marker, is identified in a first ROI, eg, a first cancerous region, and a second region phenotypic marker, eg, a second tumor marker, 9. The method of claim 8, wherein the level is identified in a second ROI, such as a second cancerous region. 9. The method of claim 8, wherein the phenotypic marker level of the first region and the phenotypic marker level of the second region, e.g., tumor marker level, are identified in the same ROI, e.g., the same cancerous region. further:
(B) identifying an ROI, eg, an ROI corresponding to a cancerous region;
The method of claim 8 comprising: 9. The method of claim 8, wherein (a) is performed before (b). 9. The method of claim 8, wherein (b) is performed before (a). Identification of the level of a first region phenotypic marker, eg, a first tumor marker, is associated with binding of a detection reagent to said first region phenotypic marker, eg, a first tumor marker, eg, proportional 9. A method according to claim 8, comprising obtaining the signal in question, e.g. directly or indirectly. 9. The method of claim 8, comprising contacting the sample with a phenotypic marker of a first region, such as a detection reagent for a first tumor marker. 9. The method of claim 8, comprising contacting the sample with an ROI marker, such as an epithelial marker detection reagent. The method of claim 8, further comprising obtaining an image of the sample and analyzing the image. 26. The method of claim 25, comprising calculating the patient's risk score from the image. 9. The method of claim 8, comprising contacting the sample with a phenotypic marker of the first region, e.g., a tumor marker detection reagent, and obtaining a binding value of the detection reagent. 28. The method of claim 27, comprising calculating the patient's risk score from the value. further:
9. The method of claim 8, comprising (b) contacting the sample with a ROI marker detection reagent. further:
30. The method of claim 29, comprising (c) defining an ROI. further:
32. The method of claim 30, comprising (d) identifying the level of a region-phenotype marker, such as a tumor marker, in the ROI. further:
32. The method of claim 31, comprising (e) analyzing the level to obtain a risk score. 33. The method of claim 32, further comprising repeating steps (a)-(d). further:
Subjecting the sample to a physical preparation step, for example,
Dissociating the sample, eg, trypsinizing, incising the sample, or contacting the sample with a detection reagent for a ROI marker;
9. The method of claim 8, comprising contacting the ROI with a detection reagent; or detecting a signal from the ROI. A method of evaluation of a patient tumor sample, e.g. a prostate tumor sample, e.g. a computer-implemented method or an automated method, comprising:
(A) identifying a first region-phenotype marker, such as a first tumor marker, or the level or amount of DNA or mRNA of said first tumor marker, such as an amount, in a ROI, such as a cancerous ROI, For example, wherein the first tumor marker is identified, selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set);
Thereby evaluating the tumor sample. The level of a first region-phenotype marker, eg, a first tumor marker from the tumor marker set, is identified in a first ROI, eg, a cancerous ROI, and a second region-phenotype marker, eg, the tumor marker 36. The method of claim 35, wherein the level of a second tumor marker from the set is identified in a second ROI, such as a second cancerous ROI. 37. The method of claim 36, wherein the first ROI, eg, a cancerous ROI and the ROI, eg, a second cancerous ROI, are identified or selected by the same method or criteria. The level of the first region-phenotype marker and the level of the second region-phenotype marker, e.g., both in the same ROI, e.g., in the same cancer, the first and second tumor marker levels from the tumor marker set 36. The method of claim 35, wherein the method is identified within a sex ROI. further:
36. The method of claim 35, comprising (b) identifying an ROI, eg, an ROI of the tumor sample corresponding to a tumor epithelium. 40. The method of claim 39, wherein (a) is performed before (b). 40. The method of claim 39, wherein (b) is performed before (a). Identification of the level of a first region-phenotype marker, eg, a first tumor marker, is associated with binding of a detection reagent to said first region-phenotype marker, eg, a first tumor marker, eg, proportional 36. The method of claim 35, comprising obtaining the signal being generated, for example, directly or indirectly. 36. The method of claim 35, wherein the tumor marker is DNA encoding FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9HSPA9. 36. The method of claim 35, wherein the tumor marker is mRNA encoding FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. 36. The method of claim 35, wherein the tumor marker is a protein selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. 36. The method of claim 35, comprising contacting the sample with a detection reagent for a marker in a tumor marker set, obtaining an image of the sample directly or indirectly, and analyzing the image. . 36. The method of claim 35, comprising calculating the patient's risk score from the image. 36. The method of claim 35, comprising contacting the sample with a detection reagent of the first marker of a tumor marker set, and obtaining a binding value of the detection reagent directly or indirectly. 49. The method of claim 48, comprising calculating a risk score for the patient from the value. further,
Identifying in a ROI (eg, the same or different ROI), eg, a ROI corresponding to a tumor epithelium, a second tumor marker selected from said tumor marker set or the DNA or mRNA level of said second tumor marker 50. A method according to any preceding claim, comprising. 51. The method of claim 50, wherein the second tumor marker is a protein of the tumor marker set. further,
Identifying a third tumor marker selected from the tumor marker set or the level of DNA or mRNA of the third tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. 51. The method of claim 50, comprising. further,
Identifying a fourth tumor marker selected from the tumor marker set or the DNA or mRNA level of the fourth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. 54. The method of claim 52, comprising. further,
Identifying a fifth tumor marker selected from the tumor marker set or a DNA or mRNA level of the fifth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. 54. The method of claim 53, comprising. further,
Identifying a sixth tumor marker selected from said tumor marker set or the DNA or mRNA level of said sixth tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium 55. The method of claim 54, comprising. further,
Identifying a seventh tumor marker selected from the tumor marker set or a DNA or mRNA level of the seventh tumor marker in an ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. 56. The method of claim 55, comprising. further,
Identifying an eighth tumor marker selected from the tumor marker set or the DNA or mRNA level of the eighth tumor marker in a ROI (eg, the same or different ROI), eg, an ROI corresponding to a tumor epithelium. 57. The method of claim 56, comprising. 58. The method of claim 57, further comprising identifying the level of additional markers disclosed herein other than those of the tumor marker set. 59. The method of claim 58, wherein the level of the additional marker is identified in a cancerous ROI. 59. The method of claim 58, wherein the level of the additional marker is identified in a benign ROI. 61. The method of any one of claims 1-60, further comprising providing the tumor sample. 61. The method of any one of claims 1-60, further comprising receiving the tumor sample from another entity, such as a hospital, laboratory, or clinic. 61. The method of any one of claims 1-60, wherein the tumor sample comprises a section or slice of prostate tissue. 61. The method of any one of claims 1-60, wherein the tumor sample comprises a plurality of portions, e.g., a plurality of prostate tissue sections or slices. 61. The method according to any one of claims 1 to 60, wherein the tumor sample is fixed, e.g. formalin fixed. 61. The method of any one of claims 1-60, wherein the tumor sample is embedded in a matrix. 61. The method of any one of claims 1-60, wherein the tumor sample is embedded in paraffin. 61. The method of any one of claims 1-60, wherein the tumor sample is deparaffinized. 61. The method according to any one of claims 1 to 60, wherein the tumor sample is a formalin-fixed paraffin embedded sample or an equivalent thereof. 70. The method of claim 69, wherein tumor sample preparation, such as deparaffinization, is automated. 71. A method according to any one of claims 1 to 70, wherein the contact between the detection reagent and the tumor sample is automated. 72. The method of any one of claims 1 to 71, wherein the tumor sample is placed in an automated scanner. 75. A portion of the tumor sample, e.g. prostate tissue, e.g. a section or slice, is placed on a substrate, e.g. on a solid or rigid substrate, e.g. a glass or plastic substrate, e.g. a glass slide. The method of any one of these. 74. A first portion of the tumor sample, such as a section or slice, is placed on a first substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. the method of. further:
Placing a second portion of the tumor sample, such as a section or slice, on a second substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. 74. The method according to 74. further:
Placing a third portion of the tumor sample, such as a section or slice, on a third substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. 75. The method according to 75. further:
Placing a fourth portion of the tumor sample, such as a section or slice, on a fourth substrate, such as a solid or rigid substrate, such as a glass or plastic substrate, such as a glass slide. 76. The method according to 76. 76. The method of claim 75, wherein the first and second portions are analyzed simultaneously. 76. The method of claim 75, wherein the first and second portions are analyzed sequentially. 80. The detection reagent of any one of claims 1 to 79, wherein the detection reagent comprises a tumor marker antibody, eg, a tumor marker monoclonal antibody, eg, a tumor marker antibody against FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. the method of. 81. The method of claim 80, wherein the tumor marker antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 82. The method of claim 81, wherein the detection reagent comprises a second antibody that is an antibody against the tumor marker antibody, such as a monoclonal antibody. 83. The method of claim 82, wherein the detection reagent comprises a third antibody that is an antibody against the second antibody, such as a monoclonal antibody. 82. The method of claim 81, wherein the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 83. The method of claim 82, wherein the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. The tumor sample:
A first ROI marker detection reagent, eg, a whole epithelial detection reagent, eg, as described herein, having a first emission profile, eg, a first peak emission, or measured in a first channel;
A second ROI marker detection reagent, eg, a basal epithelial detection reagent, eg, as described herein, having a second emission profile, eg, a second peak emission, or measured in a second channel;
Contact with a region-phenotype marker, eg, a tumor marker detection reagent, eg, one described herein, having a third emission profile, eg, a third peak emission, or measured in a third channel The method according to any one of claims 1 to 85. The tumor sample further includes:
87. The method of claim 86, wherein the method is contacted with a nuclear detection reagent having a fourth luminescence profile, e.g., a fourth peak luminescence, or measured in the fourth channel. The tumor sample further includes:
A second region-phenotype marker, eg, a second tumor marker detection reagent, eg, having a fifth luminescence profile, eg, a fifth peak emission, or measured in a fifth channel, eg, 90. The method of claim 87, wherein the method is contacted with the described. The tumor sample further includes:
A third region-phenotype marker, eg, a third tumor marker detection reagent, eg, having a sixth luminescence profile, eg, a sixth peak emission, or measured in a sixth channel, eg, herein 90. The method of claim 88, wherein the method is contacted with the described. 90. The method of any one of claims 1 to 89, wherein identifying an ROI, such as a cancerous ROI, comprises identifying a region having an epithelial structure without an outer layer of basal cells. 99. The epithelial structure is detected with a first ROI-specific detection reagent, such as a first whole epithelial specific detection reagent, such as an antibody, such as a monoclonal antibody, such as an anti-CK8 or anti-CK18 antibody, such as a monoclonal antibody. The method described in 1. An epithelial structure is detected with the first ROI-specific detection reagent, eg, the first total epithelial-specific detection reagent and the second ROI-specific detection reagent, eg, the second total epithelial-specific detection reagent, 92. The method of claim 91. The first ROI-specific detection reagent, for example, the first total epithelial-specific detection reagent and the second ROI-specific detection reagent, for example, the second total epithelial-specific detection reagent is CK8 detection reagent, 94. The method of claim 92, for example, an anti-CK8 antibody, such as a monoclonal antibody, and the other is a CK18 binding reagent, such as an anti-CK18 antibody, such as a monoclonal antibody. 94. The method of claim 93, wherein a signal of binding of the first ROI-specific detection reagent, e.g., the first whole epithelial detection reagent, is detected via a first channel, e.g., at a first wavelength. The binding signal of the first ROI-specific detection reagent, eg, the first whole epithelial detection reagent, and the signal of the second ROI-specific detection reagent, eg, the second whole epithelial detection reagent, are 94. A method according to claim 93, detected for example at a first wavelength via a channel. 92. The method of claim 91, wherein the first (and optionally second, if present) ROI-specific detection reagent, eg, the whole epithelial detection reagent, comprises a marker antibody, eg, a marker monoclonal antibody. . The first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial detection reagent, is conjugated to a label, eg, a fluorescent moiety, eg, a fluorescent dye. 92. The method according to item 91. The first (and optionally the second, if present) ROI-specific detection reagent, eg, the whole epithelial binding agent, comprises a second antibody that is an antibody against the marker antibody, eg, a monoclonal antibody 92. The method of claim 91. A third antibody wherein the first (and optionally the second, if present) ROI-specific detection reagent, eg the whole epithelial binding agent, is an antibody against the second antibody, eg a monoclonal antibody 99. The method of claim 98, comprising: 100. The method of claim 99, wherein the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 99. The method of claim 98, wherein the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 102. The presence or absence of basal cells is detected with an ROI-specific detection reagent, such as a basal epithelial detection reagent, such as the basal epithelial detection reagent described herein. Method. 105. The method of any one of claims 1-102, further comprising identifying an ROI corresponding to a benign ROI of the tumor sample, such as a second ROI. 104. The method of any one of claims 103, wherein identifying a benign ROI comprises identifying a region having an epithelial structure surrounded by an outer layer of basal cells. 105. The method of claim 104, wherein the basal cells are detected with a ROI-specific detection reagent for basal epithelium, such as an antibody, such as a monoclonal antibody, such as an anti-CK5 antibody, such as a monoclonal antibody or an anti-TRIM29 antibody, such as a monoclonal antibody. A basal cell is said ROI-specific detection reagent for the basal epithelium and a second ROI-specific detection reagent for the basal epithelium, such as an antibody, such as a monoclonal antibody, such as an anti-CK5 antibody, such as a monoclonal antibody or an anti-TRIM29 antibody, such as a monoclonal antibody. 106. The method of claim 105, wherein the method is detected. One of the first ROI-specific detection reagent of the basal epithelium and the ROI-specific detection reagent of the basal epithelium is a CK5 detection reagent, such as an anti-CK5 antibody, such as a monoclonal antibody, and the other is a TRIM29 detection reagent, such as an anti-TRIM29 antibody 107. The method of claim 106, e.g., a monoclonal antibody. 108. The method of claim 107, wherein the signal of binding of the first ROI-specific detection reagent of the basal epithelium is detected via the first channel, for example at a first wavelength. The signal of the binding of the first ROI-specific detection reagent of the basal epithelium and the signal of the second ROI-specific detection reagent of the basal epithelium are detected via the first channel, for example at a first wavelength, 109. The method of claim 108. 106. The method of claim 105, wherein the first (and optionally second, if present) ROI-specific detection reagent of the basal epithelium comprises a marker antibody, such as a marker monoclonal antibody. 106. The first (and optionally second if present) ROI-specific detection reagent of the basal epithelium is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. the method of. 106. The first (and optionally second, if present) ROI-specific detection reagent of the basal epithelium comprises a second antibody against the marker antibody, such as a monoclonal antibody, for example. Method. The first (and optionally the second, if necessary) ROI-specific detection reagent of the basal epithelium comprises a third antibody that is an antibody against the second antibody, eg, a monoclonal antibody. 105. The method according to 105. 114. The method of claim 113, wherein the second antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 119. The method of claim 114, wherein the third antibody is conjugated to a label, such as a fluorescent moiety, such as a fluorescent dye. 116. The method of any one of claims 1-115, further comprising identifying the ROI of the tumor sample as a stroma. (I.a) obtaining a signal of all epithelial specific markers, eg CK8, directly or indirectly;
117. The method of any one of claims 1-116, comprising obtaining (ii.a) a signal of a basal epithelial specific marker, such as CK5, directly or indirectly. further:
(Ib) obtaining a signal of a second whole epithelial specific marker, eg CK18, directly or indirectly;
118. The method of claim 117, comprising (ii.b) obtaining a signal of a second basal epithelial specific marker, such as TRIM29, directly or indirectly. further:
119. The method of claim 118, comprising obtaining (iii) a nuclear marker signal directly or indirectly. further;
120. The method of claim 119, comprising (iv) obtaining a signal of a second tumor marker of the tumor marker set directly or indirectly. further;
121. The method of claim 120, comprising (v) obtaining a signal of a third tumor marker of the tumor marker set directly or indirectly. further;
121. The method of claim 120, comprising (vi) obtaining a signal of a fourth tumor marker of the tumor marker set directly or indirectly. further;
129. The method of claim 122, comprising obtaining (vii) a signal of a fifth tumor marker of the tumor marker set directly or indirectly. further;
124. The method of claim 123, comprising obtaining a signal of a sixth tumor marker of the tumor marker set directly or indirectly. further;
(Ix) The method of claim 124, comprising directly or indirectly obtaining a signal of a seventh tumor marker of the tumor marker set. further;
126. The method of claim 125, comprising (x) obtaining a signal of an eighth tumor marker of the tumor marker set directly or indirectly. 119. The method of claim 118, wherein the signals of (i.a) and (ib) have the same peak emission or are collected in the same channel. 119. The method of claim 118, wherein the signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel. (I.a) obtaining a signal of all epithelial specific markers, eg CK8, directly or indirectly;
(Ib) obtaining a signal of a second whole epithelial specific marker, eg CK18, directly or indirectly;
(Ii.a) obtaining a signal of a basal epithelial specific marker, such as CK5, directly or indirectly;
(Ii.b) obtaining a signal of a second basal epithelial specific marker, such as TRIM29, directly or indirectly;
(Iii) obtaining a nuclear marker signal directly or indirectly;
(Iv) obtaining a signal of the first tumor marker directly or indirectly;
117. (v) obtaining a signal of a second tumor marker directly or indirectly; or (vi) obtaining a signal of a third tumor marker directly or indirectly. The method according to claim 1. 129. The method of claim 129, comprising (i.a), (ii.a), (iii) and (iv). 129. The method of claim 129, comprising (i.a), (ib), (ii.a), (ii.b), (iii) and (iv). 129. The method of claim 129, comprising all of (i.a)-(v). 130. The method of claim 129, comprising all of (i.a)-(vi). 134. The method of any one of claims 1-133, further comprising identifying a quality control marker level, e.g. in a second ROI, e.g. in a benign ROI. 135. The method of claim 134, further wherein the quality control marker is selected from a tumor marker set, such as DERL1. 138. The method of claim 134 or 135, further comprising contacting the sample with a detection reagent for the quality control marker. Furthermore, for example directly or indirectly obtaining a signal that is related, for example proportional, to the binding of the detection reagent to the first quality control marker, for example in a second ROI, for example in a benign ROI. 137. The method of any one of claims 133-136. 138. The method of claim 137, further comprising identifying a second quality control marker level, eg, in a second ROI, eg, in a benign ROI. 138. The method of claim 138, wherein the second quality control marker is other than a marker of the tumor marker set. 140. The method of claim 139, wherein the second quality control marker is associated with tumor lethality or aggressiveness. 139. The method of claim 138, wherein the second quality control marker is a marker described herein other than a marker of the tumor marker set, such as a tumor marker. 142. The method of claim 141, wherein the second quality control marker is selected from ACTN and VDAC1. 142. The method according to any one of claims 137 to 141, further comprising identifying the level of a third quality control marker, e.g. in a second ROI, e.g. in a benign ROI. 145. The method of claim 143, wherein the third quality control marker is other than a marker of the tumor marker set. 145. The method of claim 144, wherein the third quality control marker is a marker described herein other than a marker of the tumor marker set, such as a tumor marker. 145. The method of claim 144, wherein the third quality control marker is selected from ACTN and VDAC1. further:
Identifying the level, eg, amount, of a first quality control marker, eg, DERL1, in a second ROI, eg, benign ROI; and a second quality control marker, eg, in a benign ROI, eg, 147. The method of any one of claims 1-146, comprising identifying the level of one of ACTN and VDAC. further:
148. The method of claim 147, comprising identifying a third quality control marker in a second ROI, eg, a benign ROI, eg, one level of ACTN and VDAC. 145. The method of claim 143, wherein levels of the first, second and third quality control markers are identified in the same second ROI, eg, benign ROI. 145. The method of claim 143, wherein levels of the first, second and third quality control markers are identified in different second ROIs, eg, different benign ROIs. further:
Identifying the level of a first quality control marker, eg DERL1, in a second ROI, eg in a benign ROI;
Identifying a second quality control marker within a second ROI, eg, a benign ROI, eg, one level of ACTN and VDAC; and a third quality control within a second ROI, eg, a benign ROI 145. The method of claim 143, comprising identifying a level of one of a marker, eg, ACTN and VDAC, wherein, depending on the level, the sample is classified, for example, as not passing or not passing. 153. The method of claim 151, comprising detecting the level of a signal of one of the quality control markers. 153. The method of claim 152, wherein a first value of the detected signal indicates a first quality level, e.g., an acceptable quality, and a second value of the detected signal indicates a second quality level, e.g., an unacceptable quality. The method described. 154. The method of claim 153, wherein depending on the value, the sample is processed or not processed, eg, discarded, or analysis parameters are changed. Acquiring a multispectral image from the sample, and transferring the multispectral image to the following channels:
A first ROI-specific detection reagent, such as a channel of an epithelial-specific marker;
A second ROI-specific detection reagent, such as a channel of a basal epithelial-specific marker;
155. The method of any one of claims 1-154, comprising unmixing into a channel of a nuclear specific signal, such as a DAPI signal; and a first population phenotype marker, such as a channel of a first tumor marker. Method. Use of a first channel to collect a signal of a first ROI-specific detection reagent, eg, a whole epithelial marker;
Use of a second channel to collect signals of a second ROI-specific detection reagent, such as a basal epithelial marker;
Use of a third channel to collect nuclear region signals;
Including the use of a fourth channel to collect signals of a first tumor marker selected from a first population phenotypic marker such as FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 156. The method according to item 155. further:
Including the use of a fifth channel to collect signals of a second tumor marker selected from a second population phenotypic marker such as FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 156. The method according to item 156. further:
Including the use of a sixth channel to collect signals of a third tumor marker selected from a third population phenotypic marker such as FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 158. The method according to item 157. 159. A method according to any one of the preceding claims, comprising obtaining an image of an area of the sample to be analyzed, for example as a DAPI filtered image. 160. The method of any one of claims 1-159, comprising performing tissue localization, for example, by applying a tissue search algorithm to an image collected from the sample. 171. The method of any one of claims 1-160, comprising reacquiring an image using DAPI and FITC monochrome filters. 166. The method of any one of claims 1-161, further comprising applying a tissue search algorithm to ensure that, for example, a preselected number of fields of view including sufficient tissue is acquired. Method. 163. The method according to any one of claims 1-162, further comprising obtaining directly or indirectly continuous exposure doses of DAPI, FITC, TRITC and Cy5 filters. 164. The method of any one of claims 1-163, comprising obtaining a multispectral image of a region to be analyzed of a sample. 165. The method of any one of claims 1-164, comprising segmenting the sample region into epithelial cells, basal cells, and stroma. 166. The method of any one of claims 1-165, further comprising identifying a cytoplasmic region and a nuclear region of the sample region. 173. A method according to any one of the preceding claims comprising obtaining the value of a population phenotypic marker, such as a tumor marker, in the cytoplasm, nucleus and / or whole cell of a cancerous ROI, for example directly or indirectly . 168. A method according to any one of claims 1-167, comprising obtaining the value of a population phenotypic marker, such as a tumor marker, in the cytoplasm, nucleus and / or whole cell of benign ROI, for example directly or indirectly. 169. The method of any one of claims 1-168, wherein the tumor sample comprises a plurality of portions, such as a plurality of sections or slices. Performing the steps described herein, eg, collecting or obtaining a signal of a first population phenotypic marker, eg, a first tumor marker, from a first portion, eg, a section or slice, or an image Forming, eg, identifying a level; and performing the steps described herein, eg, from a second portion, eg, a second section or slice, from a second population phenotypic marker, eg, a second 169. The method of claim 169, comprising collecting or obtaining a signal of a tumor marker of, or forming an image, eg, identifying a level. 171. The method of claim 170, wherein the second tumor marker is selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9. further:
Identifying an ROI corresponding to a tumor epithelium in a second portion of the tumor sample, eg, a second section or slice;
Obtaining, for example directly or indirectly, a signal of a second tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 from the ROI corresponding to the tumor epithelium. 170. The method according to 170. (Ii) obtaining a signal of an epithelial specific marker, eg CK8, of said second portion of said tumor sample, eg a second section or slice;
171. The method of claim 170, comprising obtaining (ii.a) a signal of a basal epithelial specific marker, such as CK5. Further, the second portion of the tumor sample, eg, a second section or slice:
(Ib) obtaining a signal of a second epithelial specific marker, eg CK18;
178. The method of claim 173, comprising obtaining a signal of (ii.b) a second basal epithelial specific marker, such as TRIM29. Further, the second portion of the tumor sample, eg, a second section or slice:
175. The method of claim 174, comprising obtaining (iii) a nuclear marker signal. Further, of the second portion of the tumor sample, such as a second section or slice;
174. The method of claim 175, comprising obtaining a signal of the second tumor marker of claim 1. Further, of the second portion of the tumor sample, such as a second section or slice;
177. The method of claim 176, comprising obtaining (v) a signal of a second tumor marker of the tumor marker set directly or indirectly. 177. (vi) directly or indirectly obtaining a signal of a third tumor marker of the tumor marker set of the second portion of the tumor sample, eg, a second section or slice. The method described in 1. Further, of the second portion of the tumor sample, such as a second section or slice;
178. The method of claim 178, comprising directly or indirectly obtaining a signal of a fourth tumor marker of the tumor marker set. Further, of the second portion of the tumor sample, such as a second section or slice;
179. The method of claim 179, comprising obtaining a signal of a fifth tumor marker of the tumor marker set directly or indirectly. Further, of the second portion of the tumor sample, such as a second section or slice;
181. The method of claim 180, comprising obtaining a signal of a sixth tumor marker of the tumor marker set directly or indirectly. Further, of the second portion of the tumor sample, such as a second section or slice;
181. The method of claim 181, comprising directly or indirectly obtaining a signal of a seventh tumor marker of the tumor marker set. Further, of the second portion of the tumor sample, such as a second section or slice;
181. The method of claim 182 comprising directly or indirectly obtaining a signal of an eighth tumor marker of the tumor marker set. 175. The method of claim 174, wherein the signals of (i.a) and (i.b) have the same peak emission or are collected in the same channel. 175. The method of claim 174, wherein said signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel. further:
Identifying an ROI corresponding to a tumor epithelium in a third portion of the tumor sample, eg, a third section or slice;
Obtaining a signal of a third tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9, for example directly or indirectly, from the ROI corresponding to the tumor epithelium. 172. The method according to 172. Of a third portion of the tumor sample, eg, a third section or slice:
(I.a) obtaining a signal of an epithelial specific marker, eg CK8;
173. The method of claim 172, comprising obtaining (ii.a) a signal of a basal epithelial specific marker, such as CK5. Further, of the third portion of the tumor sample, eg, a third section or slice:
(Ib) obtaining a signal of a second epithelial specific marker, eg CK18;
187. The method of claim 187, comprising obtaining (ii.b) a signal of a second basal epithelial specific marker, such as TRIM29. Further, of the third portion of the tumor sample, eg, a third section or slice:
189. The method of claim 188, comprising (iii) obtaining a nuclear marker signal. Further, of the third portion of the tumor sample, eg, a third section or slice:
184. The method of claim 189, comprising obtaining a signal of the second tumor marker of claim 1. 189. The method of claim 188, wherein the signals of (i.a) and (ib) have the same peak emission or are collected in the same channel. 189. The method of claim 188, wherein the signals of (ii.a) and (ii.b) have the same peak emission or are collected in the same channel. 193. The method of any one of claims 1-192, wherein a portion of a first tumor sample, such as a first section or slice, is placed on a first substrate. 194. The method of claim 193, wherein a portion of the second tumor sample, such as a second section or slice, is placed on the second substrate. 195. The method of claim 194, wherein a portion of a third tumor sample, such as a third section or slice, is placed on a third substrate. 196. The method of claim 195, wherein a portion of a fourth tumor sample, such as a fourth section or slice, is placed on a fourth substrate. 175. The method of claim 174, wherein a portion of a first tumor sample, such as a first section or slice, and a portion of a second tumor sample, such as a second section or slice, are mounted on the same substrate. Further comprising storing or storing values corresponding to signals, values or images obtained from the sample in any step of the methods described herein in digital or electronic media, for example in a computer database, 199. A method according to any one of claims 1-197. 199. The method of any one of claims 1-198, comprising exporting a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, to identify a nuclear region. The method described. 200. The method of any one of claims 1-199, comprising exporting a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, to identify a cytoplasmic region. The method described. 200. Any one of claims 1 to 200, comprising exporting a value or image obtained by capturing a signal from the tumor sample to software, eg, pattern recognition or object recognition software, to identify a cancerous ROI. The method described in 1. 202. The value or image obtained by capturing a signal from the tumor sample is exported to software, eg, pattern recognition or object recognition software, to identify a benign ROI. The method described. Exporting values or images obtained by capturing signals from the tumor sample to software, e.g., pattern recognition or object recognition software, to obtain a value for the level of the tumor marker in a cancerous ROI. The method according to any one of 1 to 202. 2. Exporting values or images obtained by capturing signals from the tumor sample to software, e.g., pattern recognition or object recognition software, to obtain a value for the level of the tumor marker in a benign ROI. The method of any one of -203. In response to a phenotypic marker of the region, such as a signal of a tumor marker, a signal of a first ROI marker, such as a whole epithelial specific marker and a signal of a second ROI marker, such as a basal epithelial specific marker, 205. A method according to any one of claims 1 to 204, comprising obtaining a value for the level of a phenotypic marker of a region, such as a tumor marker. Tumors within a benign ROI in response to a phenotypic marker of the region, such as a signal of a tumor marker, a first ROI marker, such as a signal of a whole epithelial specific marker and a signal of a second ROI marker, such as a basal epithelial specific marker 207. The method of any one of claims 1-205, comprising obtaining a value for a marker level. Regional phenotypic markers such as tumor marker signals, first ROI markers such as whole epithelial specific marker signals and second ROI markers such as basal epithelial specific marker signals and third ROI markers such as nuclei 207. The method of any one of claims 1-206, comprising obtaining a cytoplasmic level value of a tumor marker within a cancerous ROI in response to a specific marker signal. Regional phenotypic markers such as tumor marker signals, first ROI markers such as whole epithelial specific marker signals and second ROI markers such as basal epithelial specific marker signals and third ROI markers such as nuclei 207. The method of any one of claims 1 to 207, comprising obtaining a value for a nuclear level of a tumor marker within a benign ROI in response to a signal of a specific marker. 206. The method of claim 205, comprising calculating a risk score for the patient. 206. The method of claim 205, comprising calculating the patient's risk score in response to the value. 211. The method of claim 210, comprising calculating a risk score for the patient in response to one or more of the values of claims 198-208. 213. The method of claim 210, comprising calculating a risk score for the patient, wherein the risk score correlates with the likelihood of extraprostatic expansion or metastasis. Performing a prognosis of the patient according to the risk score, classifying the patient, selecting a course of treatment for the patient, or applying a selected course of treatment to the patient. 223. The method of claim 212. 223. The method of claim 211, wherein the risk score corresponds to a “good prognosis” case (eg, surgical Gleason 3 + 3 or 3 with a minimum of 4, with organ-localized (≦ T2) tumor). 223. The method of claim 211, wherein the risk score corresponds to a “poor prognosis” case (eg, capsule penetration (T3a), seminal vesicle invasion (T3b), lymph node metastasis, or a dominant Gleason pattern of 4 or higher). . According to the risk score, “good prognosis” cases (eg surgical Gleason 3 + 3 or 3 with a minimum of 4 with organ-localized (≦ T2) tumor) and “poor prognosis” cases (eg capsule penetration (T3a), seminal vesicles 223. The method of claim 211, wherein discrimination between invasion (T3b), lymph node metastasis, or a dominant Gleason pattern of 4 or higher is possible. The risk score is:
Surgical Gleason 3 + 3 or localized disease (≦ T3a) (defined as “low risk”);
Surgical Gleason ≧ 3 + 4 or non-localized disease (T3b, N or M) (defined as “medium to high risk”);
Surgical Gleason ≦ 3 + 4 and organ-localized disease (≦ T2) (defined as “good prognosis”); or Surgical Gleason ≧ 4 + 3 or non-organ-localized disease (T3a, T3b, N or M) (“poor prognosis”)
221. The method of claim 211, wherein the method corresponds to or predicts. 211. The method of claim 211, further comprising identifying the patient as having an aggressive cancer or having an increased risk or cancer-related lethal outcome as a function of the risk score. 211. The method of claim 211, comprising selecting the patient for adjuvant therapy or administering the patient to adjuvant therapy. A kit comprising the detection reagents 1, 2, 3, 4, 5, 6, 7 or all of the tumor markers FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9. The kit according to claim 220, further comprising detection reagents for whole epithelial markers and basal epithelial markers. On top of a cancer sample, eg a prostate tumor sample:
Detection reagent for all epithelial markers;
Detection reagent for basal epithelial marker;
A cancer sample, for example, a prostate tumor sample, on which a detection reagent for a tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 or HSPA9 is arranged. 223. A prostate tumor sample according to claim 222, wherein the sample comprises a plurality of portions, such as slices. A cancer sample, for example, a prostate tumor sample, wherein a detection reagent for a second tumor marker selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2, or HSPA9 is further disposed on the cancer sample, for example, a prostate tumor sample . A computer-implemented method for evaluation of a patient's prostate tumor sample, comprising:
(I) identifying the ROI (cancerous ROI) of the tumor sample corresponding to the tumor epithelium;
(Ii) identifying the level, eg, amount, of each of the following tumor markers within the cancerous ROI: FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9 (tumor marker set), where Identifying a tumor marker level comprising identifying, for example, directly or indirectly, a signal that is related, eg, proportional, to the binding of an antibody to said tumor marker;
(Iii) obtaining a value for each level of the tumor marker within the cancerous ROI; and (iv) assigning a risk score to the patient by combining the levels, eg, by an algorithm, depending on the value Evaluating the tumor sample, thereby evaluating a prostate tumor sample. Use of a first channel to collect signals of all epithelial markers;
Use of a second channel to collect basal epithelial marker signals;
Use of a third channel to collect nuclear region signals;
226. The method of claim 225, comprising the use of a fourth channel to collect a tumor marker signal selected from FUS, SMAD4, DERL1, YBX1, pS6, PDSS2, CUL2 and HSPA9. 225. The level of the first tumor marker of the tumor marker set is identified in a first cancerous ROI, and the level of a second tumor marker of the tumor marker set is identified in a second cancerous ROI. 226. A method according to any of 226. 228. The method of any of claims 225-227, wherein both the level of the first tumor marker and the level of the second tumor marker from the tumor marker set are identified in the same cancerous ROI. further:
Identifying the level of a first quality control marker, eg DERL1, in a second ROI, eg in a benign ROI;
Identifying a second quality control marker within a second ROI, eg, a benign ROI, eg, one level of ACTN and VDAC; and a third quality control within a second ROI, eg, a benign ROI 229. Identifying a level of one of a marker, eg, ACTN and VDAC, wherein, according to the level, comprising classifying the sample, for example, not passing or not passing, according to any of claims 225-228. the method of.